mirror of
git://sourceware.org/git/lvm2.git
synced 2024-12-29 15:22:30 +03:00
6552 lines
184 KiB
C
6552 lines
184 KiB
C
/*
|
|
* Copyright (C) 2005-2023 Red Hat, Inc. All rights reserved.
|
|
*
|
|
* This file is part of LVM2.
|
|
*
|
|
* This copyrighted material is made available to anyone wishing to use,
|
|
* modify, copy, or redistribute it subject to the terms and conditions
|
|
* of the GNU Lesser General Public License v.2.1.
|
|
*
|
|
* You should have received a copy of the GNU Lesser General Public License
|
|
* along with this program; if not, write to the Free Software Foundation,
|
|
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|
*/
|
|
|
|
#include "tools.h"
|
|
|
|
#include "lib/lvmpolld/polldaemon.h"
|
|
#include "lib/metadata/lv_alloc.h"
|
|
#include "lib/metadata/metadata.h"
|
|
#include "lvconvert_poll.h"
|
|
|
|
typedef enum {
|
|
/* Split:
|
|
* For a mirrored or raid LV, split mirror into two mirrors, optionally tracking
|
|
* future changes to the main mirror to allow future recombination.
|
|
*/
|
|
CONV_SPLIT_MIRRORS = 2,
|
|
|
|
/* Every other segment type or mirror log conversion we haven't separated out */
|
|
CONV_OTHER = 3,
|
|
} conversion_type_t;
|
|
|
|
struct lvconvert_params {
|
|
/* Exactly one of these 12 command categories is determined */
|
|
int keep_mimages; /* 2 */ /* --splitmirrors */
|
|
/* other */ /* 3 */
|
|
|
|
/* FIXME Eliminate all cases where more than one of the above are set then use conv_type instead */
|
|
conversion_type_t conv_type;
|
|
|
|
int track_changes; /* CONV_SPLIT_MIRRORS is set */
|
|
|
|
int corelog; /* Equivalent to --mirrorlog core */
|
|
int mirrorlog; /* Only one of corelog and mirrorlog may be set */
|
|
|
|
int mirrors_supplied; /* When type_str is not set, this may be set with keep_mimages for --splitmirrors */
|
|
const char *type_str; /* When this is set, mirrors_supplied may optionally also be set */
|
|
/* Holds what you asked for based on --type or other arguments, else "" */
|
|
|
|
const struct segment_type *segtype; /* Holds what segment type you will get */
|
|
|
|
int force;
|
|
int yes;
|
|
int zero;
|
|
|
|
const char *lv_name;
|
|
const char *lv_split_name;
|
|
const char *lv_name_full;
|
|
const char *vg_name;
|
|
int wait_completion;
|
|
int need_polling;
|
|
|
|
uint32_t region_size;
|
|
unsigned region_size_supplied;
|
|
|
|
uint32_t mirrors;
|
|
sign_t mirrors_sign;
|
|
uint32_t stripes;
|
|
uint32_t stripe_size;
|
|
unsigned stripes_supplied;
|
|
unsigned stripe_size_supplied;
|
|
uint32_t read_ahead;
|
|
|
|
unsigned target_attr;
|
|
|
|
alloc_policy_t alloc;
|
|
|
|
int pv_count;
|
|
char **pvs;
|
|
struct dm_list *pvh;
|
|
|
|
struct logical_volume *lv_to_poll;
|
|
struct dm_list idls;
|
|
|
|
const char *origin_name;
|
|
};
|
|
|
|
struct convert_poll_id_list {
|
|
struct dm_list list;
|
|
struct poll_operation_id *id;
|
|
unsigned is_merging_origin:1;
|
|
unsigned is_merging_origin_thin:1;
|
|
};
|
|
|
|
/* FIXME Temporary function until the enum replaces the separate variables */
|
|
static void _set_conv_type(struct lvconvert_params *lp, conversion_type_t conv_type)
|
|
{
|
|
if (lp->conv_type != CONV_OTHER)
|
|
log_error(INTERNAL_ERROR "Changing conv_type from %d to %d.", lp->conv_type, conv_type);
|
|
|
|
lp->conv_type = conv_type;
|
|
}
|
|
|
|
static int _raid0_type_requested(const char *type_str)
|
|
{
|
|
return (!strcmp(type_str, SEG_TYPE_NAME_RAID0) || !strcmp(type_str, SEG_TYPE_NAME_RAID0_META));
|
|
}
|
|
|
|
/* mirror/raid* (1,10,4,5,6 and their variants) reshape */
|
|
static int _mirror_or_raid_type_requested(struct cmd_context *cmd, const char *type_str)
|
|
{
|
|
return (arg_is_set(cmd, mirrors_ARG) || !strcmp(type_str, SEG_TYPE_NAME_MIRROR) ||
|
|
(!strncmp(type_str, SEG_TYPE_NAME_RAID, 4) && !_raid0_type_requested(type_str)));
|
|
}
|
|
|
|
static int _linear_type_requested(const char *type_str)
|
|
{
|
|
return (!strcmp(type_str, SEG_TYPE_NAME_LINEAR));
|
|
}
|
|
|
|
static int _striped_type_requested(const char *type_str)
|
|
{
|
|
return (!strcmp(type_str, SEG_TYPE_NAME_STRIPED) || _linear_type_requested(type_str));
|
|
}
|
|
|
|
static int _read_conversion_type(struct cmd_context *cmd,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
|
|
const char *type_str = arg_str_value(cmd, type_ARG, "");
|
|
|
|
lp->type_str = type_str;
|
|
if (!lp->type_str[0])
|
|
return 1;
|
|
|
|
/* FIXME: Check thin-pool and thin more thoroughly! */
|
|
if (!strcmp(type_str, SEG_TYPE_NAME_SNAPSHOT) || _striped_type_requested(type_str) ||
|
|
!strncmp(type_str, SEG_TYPE_NAME_RAID, 4) || !strcmp(type_str, SEG_TYPE_NAME_MIRROR) ||
|
|
!strcmp(type_str, SEG_TYPE_NAME_CACHE_POOL) || !strcmp(type_str, SEG_TYPE_NAME_CACHE) ||
|
|
!strcmp(type_str, SEG_TYPE_NAME_THIN_POOL) || !strcmp(type_str, SEG_TYPE_NAME_THIN))
|
|
return 1;
|
|
|
|
log_error("Conversion using --type %s is not supported.", type_str);
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int _read_params(struct cmd_context *cmd, struct lvconvert_params *lp)
|
|
{
|
|
const char *vg_name = NULL;
|
|
|
|
if (!_read_conversion_type(cmd, lp))
|
|
return_0;
|
|
|
|
if (!arg_is_set(cmd, background_ARG))
|
|
lp->wait_completion = 1;
|
|
|
|
if (arg_is_set(cmd, corelog_ARG))
|
|
lp->corelog = 1;
|
|
|
|
if (arg_is_set(cmd, mirrorlog_ARG)) {
|
|
if (lp->corelog) {
|
|
log_error("--mirrorlog and --corelog are incompatible.");
|
|
return 0;
|
|
}
|
|
lp->mirrorlog = 1;
|
|
}
|
|
|
|
if (arg_is_set(cmd, trackchanges_ARG))
|
|
lp->track_changes = 1;
|
|
|
|
/*
|
|
* The '--splitmirrors n' argument is equivalent to '--mirrors -n'
|
|
* (note the minus sign), except that it signifies the additional
|
|
* intent to keep the mimage that is detached, rather than
|
|
* discarding it.
|
|
*/
|
|
if (arg_is_set(cmd, splitmirrors_ARG)) {
|
|
if ((lp->lv_split_name = arg_str_value(cmd, name_ARG, NULL))) {
|
|
if (!validate_restricted_lvname_param(cmd, &vg_name, &lp->lv_split_name))
|
|
return_0;
|
|
}
|
|
|
|
if (_mirror_or_raid_type_requested(cmd, lp->type_str)) {
|
|
log_error("--mirrors/--type mirror/--type raid* and --splitmirrors are "
|
|
"mutually exclusive.");
|
|
return 0;
|
|
}
|
|
|
|
if (!arg_is_set(cmd, name_ARG) && !lp->track_changes) {
|
|
log_error("Please name the new logical volume using '--name'");
|
|
return 0;
|
|
}
|
|
|
|
if ((lp->lv_split_name = arg_str_value(cmd, name_ARG, NULL))) {
|
|
if (!validate_restricted_lvname_param(cmd, &vg_name, &lp->lv_split_name))
|
|
return_0;
|
|
}
|
|
|
|
lp->keep_mimages = 1;
|
|
_set_conv_type(lp, CONV_SPLIT_MIRRORS);
|
|
lp->mirrors = arg_uint_value(cmd, splitmirrors_ARG, 0);
|
|
lp->mirrors_sign = SIGN_MINUS;
|
|
}
|
|
|
|
/* If no other case was identified, then use of --stripes means --type striped */
|
|
if (!arg_is_set(cmd, type_ARG) && !*lp->type_str &&
|
|
!lp->mirrorlog && !lp->corelog &&
|
|
(arg_is_set(cmd, stripes_long_ARG) || arg_is_set(cmd, stripesize_ARG)))
|
|
lp->type_str = SEG_TYPE_NAME_STRIPED;
|
|
|
|
if ((arg_is_set(cmd, stripes_long_ARG) || arg_is_set(cmd, stripesize_ARG)) &&
|
|
!(_mirror_or_raid_type_requested(cmd, lp->type_str) || _striped_type_requested(lp->type_str) ||
|
|
_raid0_type_requested(lp->type_str) || arg_is_set(cmd, thinpool_ARG))) {
|
|
log_error("--stripes or --stripesize argument is only valid "
|
|
"with --mirrors/--type mirror/--type raid*/--type striped/--type linear, --repair and --thinpool");
|
|
return 0;
|
|
}
|
|
|
|
if (arg_is_set(cmd, mirrors_ARG)) {
|
|
/* --splitmirrors is the mechanism for detaching and keeping a mimage */
|
|
lp->mirrors_supplied = 1;
|
|
lp->mirrors = arg_uint_value(cmd, mirrors_ARG, 0);
|
|
lp->mirrors_sign = arg_sign_value(cmd, mirrors_ARG, SIGN_NONE);
|
|
}
|
|
|
|
lp->alloc = (alloc_policy_t) arg_uint_value(cmd, alloc_ARG, ALLOC_INHERIT);
|
|
|
|
/*
|
|
* Final checking of each case:
|
|
* lp->keep_mimages
|
|
* --type mirror|raid lp->mirrorlog lp->corelog
|
|
* --type raid0|striped
|
|
*/
|
|
switch(lp->conv_type) {
|
|
case CONV_SPLIT_MIRRORS:
|
|
break;
|
|
|
|
case CONV_OTHER:
|
|
if (arg_is_set(cmd, regionsize_ARG)) {
|
|
lp->region_size = arg_uint_value(cmd, regionsize_ARG, 0);
|
|
lp->region_size_supplied = 1;
|
|
} else {
|
|
lp->region_size = get_default_region_size(cmd);
|
|
lp->region_size_supplied = 0;
|
|
}
|
|
|
|
if (_mirror_or_raid_type_requested(cmd, lp->type_str) ||
|
|
lp->mirrorlog || lp->corelog) { /* Mirrors (and some RAID functions) */
|
|
if (arg_is_set(cmd, chunksize_ARG)) {
|
|
log_error("--chunksize is only available with snapshots or pools.");
|
|
return 0;
|
|
}
|
|
|
|
if (arg_is_set(cmd, zero_ARG)) {
|
|
log_error("--zero is only available with snapshots or pools.");
|
|
return 0;
|
|
}
|
|
|
|
/* FIXME man page says in one place that --type and --mirrors can't be mixed */
|
|
if (lp->mirrors_supplied && !lp->mirrors)
|
|
/* down-converting to linear/stripe? */
|
|
lp->type_str = SEG_TYPE_NAME_STRIPED;
|
|
|
|
} else if (_raid0_type_requested(lp->type_str) || _striped_type_requested(lp->type_str)) { /* striped or linear or raid0 */
|
|
if (arg_from_list_is_set(cmd, "cannot be used with --type raid0 or --type striped or --type linear",
|
|
chunksize_ARG, corelog_ARG, mirrors_ARG, mirrorlog_ARG, zero_ARG,
|
|
-1))
|
|
return_0;
|
|
} /* else segtype will default to current type */
|
|
}
|
|
|
|
lp->force = arg_count(cmd, force_ARG);
|
|
lp->yes = arg_count(cmd, yes_ARG);
|
|
|
|
return 1;
|
|
}
|
|
|
|
|
|
static const struct poll_functions _lvconvert_mirror_fns = {
|
|
.poll_progress = poll_mirror_progress,
|
|
.finish_copy = lvconvert_mirror_finish,
|
|
};
|
|
|
|
static const struct poll_functions _lvconvert_merge_fns = {
|
|
.poll_progress = poll_merge_progress,
|
|
.finish_copy = lvconvert_merge_finish,
|
|
};
|
|
|
|
static const struct poll_functions _lvconvert_thin_merge_fns = {
|
|
.poll_progress = poll_thin_merge_progress,
|
|
.finish_copy = lvconvert_merge_finish,
|
|
};
|
|
|
|
static struct poll_operation_id *_create_id(struct cmd_context *cmd,
|
|
const char *vg_name,
|
|
const char *lv_name,
|
|
const char *uuid)
|
|
{
|
|
struct poll_operation_id *id;
|
|
char lv_full_name[NAME_LEN];
|
|
|
|
if (!vg_name || !lv_name || !uuid) {
|
|
log_error(INTERNAL_ERROR "Wrong params for lvconvert _create_id.");
|
|
return NULL;
|
|
}
|
|
|
|
if (dm_snprintf(lv_full_name, sizeof(lv_full_name), "%s/%s", vg_name, lv_name) < 0) {
|
|
log_error(INTERNAL_ERROR "Name \"%s/%s\" is too long.", vg_name, lv_name);
|
|
return NULL;
|
|
}
|
|
|
|
if (!(id = dm_pool_alloc(cmd->mem, sizeof(*id)))) {
|
|
log_error("Poll operation ID allocation failed.");
|
|
return NULL;
|
|
}
|
|
|
|
if (!(id->display_name = dm_pool_strdup(cmd->mem, lv_full_name)) ||
|
|
!(id->lv_name = strchr(id->display_name, '/')) ||
|
|
!(id->vg_name = dm_pool_strdup(cmd->mem, vg_name)) ||
|
|
!(id->uuid = dm_pool_strdup(cmd->mem, uuid))) {
|
|
log_error("Failed to copy one or more poll operation ID members.");
|
|
dm_pool_free(cmd->mem, id);
|
|
return NULL;
|
|
}
|
|
|
|
id->lv_name++; /* skip over '/' */
|
|
|
|
return id;
|
|
}
|
|
|
|
static int _lvconvert_poll_by_id(struct cmd_context *cmd, struct poll_operation_id *id,
|
|
unsigned background,
|
|
int is_merging_origin,
|
|
int is_merging_origin_thin)
|
|
{
|
|
if (test_mode())
|
|
return ECMD_PROCESSED;
|
|
|
|
if (is_merging_origin)
|
|
return poll_daemon(cmd, background,
|
|
(MERGING | (is_merging_origin_thin ? THIN_VOLUME : SNAPSHOT)),
|
|
is_merging_origin_thin ? &_lvconvert_thin_merge_fns : &_lvconvert_merge_fns,
|
|
"Merged", id);
|
|
|
|
return poll_daemon(cmd, background, CONVERTING,
|
|
&_lvconvert_mirror_fns, "Converted", id);
|
|
}
|
|
|
|
int lvconvert_poll(struct cmd_context *cmd, struct logical_volume *lv,
|
|
unsigned background)
|
|
{
|
|
int r;
|
|
struct poll_operation_id *id = _create_id(cmd, lv->vg->name, lv->name, lv->lvid.s);
|
|
int is_merging_origin = 0;
|
|
int is_merging_origin_thin = 0;
|
|
|
|
if (!id) {
|
|
log_error("Failed to allocate poll identifier for lvconvert.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
/* FIXME: check this in polling instead */
|
|
if (lv_is_merging_origin(lv)) {
|
|
is_merging_origin = 1;
|
|
is_merging_origin_thin = seg_is_thin_volume(find_snapshot(lv));
|
|
}
|
|
|
|
r = _lvconvert_poll_by_id(cmd, id, background, is_merging_origin, is_merging_origin_thin);
|
|
|
|
return r;
|
|
}
|
|
|
|
static int _insert_lvconvert_layer(struct cmd_context *cmd,
|
|
struct logical_volume *lv)
|
|
{
|
|
char format[NAME_LEN], layer_name[NAME_LEN];
|
|
int i;
|
|
|
|
/*
|
|
* We would like to give the same number for this layer
|
|
* and the newly added mimage.
|
|
* However, LV name of newly added mimage is determined *after*
|
|
* the LV name of this layer is determined.
|
|
*
|
|
* So, use generate_lv_name() to generate mimage name first
|
|
* and take the number from it.
|
|
*/
|
|
|
|
if (dm_snprintf(format, sizeof(format), "%s_mimage_%%d", lv->name) < 0) {
|
|
log_error("lvconvert: layer name creation failed.");
|
|
return 0;
|
|
}
|
|
|
|
if (!generate_lv_name(lv->vg, format, layer_name, sizeof(layer_name)) ||
|
|
sscanf(layer_name, format, &i) != 1) {
|
|
log_error("lvconvert: layer name generation failed.");
|
|
return 0;
|
|
}
|
|
|
|
if (dm_snprintf(layer_name, sizeof(layer_name), MIRROR_SYNC_LAYER "_%d", i) < 0) {
|
|
log_error("layer name creation failed.");
|
|
return 0;
|
|
}
|
|
|
|
if (!insert_layer_for_lv(cmd, lv, 0, layer_name)) {
|
|
log_error("Failed to insert resync layer");
|
|
return 0;
|
|
}
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _failed_mirrors_count(struct logical_volume *lv)
|
|
{
|
|
struct lv_segment *lvseg;
|
|
int ret = 0;
|
|
unsigned s;
|
|
|
|
dm_list_iterate_items(lvseg, &lv->segments) {
|
|
if (!seg_is_mirrored(lvseg))
|
|
return -1;
|
|
for (s = 0; s < lvseg->area_count; s++) {
|
|
if (seg_type(lvseg, s) == AREA_LV) {
|
|
if (is_temporary_mirror_layer(seg_lv(lvseg, s)))
|
|
ret += _failed_mirrors_count(seg_lv(lvseg, s));
|
|
else if (lv_is_partial(seg_lv(lvseg, s)))
|
|
++ ret;
|
|
}
|
|
else if (seg_type(lvseg, s) == AREA_PV &&
|
|
is_missing_pv(seg_pv(lvseg, s)))
|
|
++ret;
|
|
}
|
|
}
|
|
|
|
return ret;
|
|
}
|
|
|
|
static int _failed_logs_count(struct logical_volume *lv)
|
|
{
|
|
int ret = 0;
|
|
unsigned s;
|
|
struct logical_volume *log_lv = first_seg(lv)->log_lv;
|
|
if (log_lv && lv_is_partial(log_lv)) {
|
|
if (lv_is_mirrored(log_lv))
|
|
ret += _failed_mirrors_count(log_lv);
|
|
else
|
|
ret += 1;
|
|
}
|
|
for (s = 0; s < first_seg(lv)->area_count; s++) {
|
|
if (seg_type(first_seg(lv), s) == AREA_LV &&
|
|
is_temporary_mirror_layer(seg_lv(first_seg(lv), s)))
|
|
ret += _failed_logs_count(seg_lv(first_seg(lv), s));
|
|
}
|
|
return ret;
|
|
}
|
|
|
|
|
|
static struct dm_list *_failed_pv_list(struct volume_group *vg)
|
|
{
|
|
struct dm_list *failed_pvs;
|
|
struct pv_list *pvl, *new_pvl;
|
|
|
|
if (!(failed_pvs = dm_pool_alloc(vg->vgmem, sizeof(*failed_pvs)))) {
|
|
log_error("Allocation of list of failed_pvs failed.");
|
|
return NULL;
|
|
}
|
|
|
|
dm_list_init(failed_pvs);
|
|
|
|
dm_list_iterate_items(pvl, &vg->pvs) {
|
|
if (!is_missing_pv(pvl->pv))
|
|
continue;
|
|
|
|
/*
|
|
* Finally, --repair will remove empty PVs.
|
|
* But we only want remove these which are output of repair,
|
|
* Do not count these which are already empty here.
|
|
* FIXME: code should traverse PV in LV not in whole VG.
|
|
* FIXME: layer violation? should it depend on vgreduce --removemising?
|
|
*/
|
|
if (pvl->pv->pe_alloc_count == 0)
|
|
continue;
|
|
|
|
if (!(new_pvl = dm_pool_zalloc(vg->vgmem, sizeof(*new_pvl)))) {
|
|
log_error("Allocation of failed_pvs list entry failed.");
|
|
return NULL;
|
|
}
|
|
new_pvl->pv = pvl->pv;
|
|
dm_list_add(failed_pvs, &new_pvl->list);
|
|
}
|
|
|
|
return failed_pvs;
|
|
}
|
|
|
|
static int _is_partial_lv(struct logical_volume *lv,
|
|
void *baton __attribute__((unused)))
|
|
{
|
|
return lv_is_partial(lv);
|
|
}
|
|
|
|
/*
|
|
* Walk down the stacked mirror LV to the original mirror LV.
|
|
*/
|
|
static struct logical_volume *_original_lv(struct logical_volume *lv)
|
|
{
|
|
struct logical_volume *next_lv = lv, *tmp_lv;
|
|
|
|
while ((tmp_lv = find_temporary_mirror(next_lv)))
|
|
next_lv = tmp_lv;
|
|
|
|
return next_lv;
|
|
}
|
|
|
|
static void _lvconvert_mirrors_repair_ask(struct cmd_context *cmd,
|
|
int failed_log, int failed_mirrors,
|
|
int *replace_log, int *replace_mirrors)
|
|
{
|
|
const char *leg_policy, *log_policy;
|
|
int force = arg_count(cmd, force_ARG);
|
|
int yes = arg_count(cmd, yes_ARG);
|
|
|
|
if (arg_is_set(cmd, usepolicies_ARG)) {
|
|
leg_policy = find_config_tree_str(cmd, activation_mirror_image_fault_policy_CFG, NULL);
|
|
log_policy = find_config_tree_str(cmd, activation_mirror_log_fault_policy_CFG, NULL);
|
|
*replace_mirrors = strcmp(leg_policy, "remove");
|
|
*replace_log = strcmp(log_policy, "remove");
|
|
return;
|
|
}
|
|
|
|
if (force != PROMPT) {
|
|
*replace_log = *replace_mirrors = 0;
|
|
return;
|
|
}
|
|
|
|
*replace_log = *replace_mirrors = 1;
|
|
|
|
if (yes)
|
|
return;
|
|
|
|
if (failed_log &&
|
|
yes_no_prompt("Attempt to replace failed mirror log? [y/n]: ") == 'n')
|
|
*replace_log = 0;
|
|
|
|
if (failed_mirrors &&
|
|
yes_no_prompt("Attempt to replace failed mirror images "
|
|
"(requires full device resync)? [y/n]: ") == 'n')
|
|
*replace_mirrors = 0;
|
|
}
|
|
|
|
/*
|
|
* _get_log_count
|
|
* @lv: the mirror LV
|
|
*
|
|
* Get the number of on-disk copies of the log.
|
|
* 0 = 'core'
|
|
* 1 = 'disk'
|
|
* 2+ = 'mirrored'
|
|
*/
|
|
static uint32_t _get_log_count(struct logical_volume *lv)
|
|
{
|
|
struct logical_volume *log_lv;
|
|
|
|
log_lv = first_seg(_original_lv(lv))->log_lv;
|
|
if (log_lv)
|
|
return lv_mirror_count(log_lv);
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int _lv_update_mirrored_log(struct logical_volume *lv,
|
|
struct dm_list *operable_pvs,
|
|
int log_count)
|
|
{
|
|
int old_log_count;
|
|
struct logical_volume *log_lv;
|
|
|
|
/*
|
|
* When log_count is 0, mirrored log doesn't need to be
|
|
* updated here but it will be removed later.
|
|
*/
|
|
if (!log_count)
|
|
return 1;
|
|
|
|
log_lv = first_seg(_original_lv(lv))->log_lv;
|
|
if (!log_lv || !lv_is_mirrored(log_lv))
|
|
return 1;
|
|
|
|
old_log_count = _get_log_count(lv);
|
|
if (old_log_count == log_count)
|
|
return 1;
|
|
|
|
/* Reducing redundancy of the log */
|
|
return remove_mirror_images(log_lv, log_count,
|
|
is_mirror_image_removable,
|
|
operable_pvs, 0U);
|
|
}
|
|
|
|
static int _lv_update_log_type(struct cmd_context *cmd,
|
|
struct lvconvert_params *lp,
|
|
struct logical_volume *lv,
|
|
struct dm_list *operable_pvs,
|
|
int log_count)
|
|
{
|
|
int old_log_count;
|
|
uint32_t region_size = (lp) ? lp->region_size :
|
|
first_seg(lv)->region_size;
|
|
alloc_policy_t alloc = (lp) ? lp->alloc : lv->alloc;
|
|
struct logical_volume *original_lv;
|
|
struct logical_volume *log_lv;
|
|
|
|
old_log_count = _get_log_count(lv);
|
|
if (old_log_count == log_count)
|
|
return 1;
|
|
|
|
original_lv = _original_lv(lv);
|
|
/* Remove an existing log completely */
|
|
if (!log_count) {
|
|
if (!remove_mirror_log(cmd, original_lv, operable_pvs,
|
|
arg_count(cmd, yes_ARG) ||
|
|
arg_count(cmd, force_ARG)))
|
|
return_0;
|
|
return 1;
|
|
}
|
|
|
|
log_lv = first_seg(original_lv)->log_lv;
|
|
|
|
/* Adding redundancy to the log */
|
|
if (old_log_count < log_count) {
|
|
if (!(region_size = adjusted_mirror_region_size(cmd, lv->vg->extent_size,
|
|
lv->le_count,
|
|
region_size, 0,
|
|
vg_is_clustered(lv->vg))))
|
|
return_0;
|
|
|
|
if (!add_mirror_log(cmd, original_lv, log_count,
|
|
region_size, operable_pvs, alloc))
|
|
return_0;
|
|
/*
|
|
* FIXME: This simple approach won't work in cluster mirrors,
|
|
* but it doesn't matter because we don't support
|
|
* mirrored logs in cluster mirrors.
|
|
*/
|
|
if (old_log_count &&
|
|
!lv_update_and_reload(log_lv))
|
|
return_0;
|
|
|
|
return 1;
|
|
}
|
|
|
|
/* Reducing redundancy of the log */
|
|
return remove_mirror_images(log_lv, log_count,
|
|
is_mirror_image_removable, operable_pvs, 1U);
|
|
}
|
|
|
|
/*
|
|
* Reomove missing and empty PVs from VG, if are also in provided list
|
|
*/
|
|
static void _remove_missing_empty_pv(struct volume_group *vg, struct dm_list *remove_pvs)
|
|
{
|
|
struct pv_list *pvl, *pvl_vg, *pvlt;
|
|
int removed = 0;
|
|
|
|
if (!remove_pvs)
|
|
return;
|
|
|
|
dm_list_iterate_items(pvl, remove_pvs) {
|
|
dm_list_iterate_items_safe(pvl_vg, pvlt, &vg->pvs) {
|
|
if (!id_equal(&pvl->pv->id, &pvl_vg->pv->id) ||
|
|
!is_missing_pv(pvl_vg->pv) ||
|
|
pvl_vg->pv->pe_alloc_count != 0)
|
|
continue;
|
|
|
|
/* FIXME: duplication of vgreduce code, move this to library */
|
|
vg->free_count -= pvl_vg->pv->pe_count;
|
|
vg->extent_count -= pvl_vg->pv->pe_count;
|
|
del_pvl_from_vgs(vg, pvl_vg);
|
|
free_pv_fid(pvl_vg->pv);
|
|
|
|
removed++;
|
|
}
|
|
}
|
|
|
|
if (removed) {
|
|
if (!vg_write(vg) || !vg_commit(vg)) {
|
|
stack;
|
|
return;
|
|
}
|
|
log_warn("WARNING: %d missing and now unallocated Physical Volumes removed from VG.", removed);
|
|
}
|
|
}
|
|
|
|
/*
|
|
* _lvconvert_mirrors_parse_params
|
|
*
|
|
* This function performs the following:
|
|
* 1) Gets the old values of mimage and log counts
|
|
* 2) Parses the CLI args to find the new desired values
|
|
* 3) Adjusts 'lp->mirrors' to the appropriate absolute value.
|
|
* (Remember, 'lp->mirrors' is specified in terms of the number of "copies"
|
|
* vs. the number of mimages. It can also be a relative value.)
|
|
* 4) Sets 'lp->need_polling' if collapsing
|
|
* 5) Validates other mirror params
|
|
*
|
|
* Returns: 1 on success, 0 on error
|
|
*/
|
|
static int _lvconvert_mirrors_parse_params(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct lvconvert_params *lp,
|
|
uint32_t *old_mimage_count,
|
|
uint32_t *old_log_count,
|
|
uint32_t *new_mimage_count,
|
|
uint32_t *new_log_count)
|
|
{
|
|
*old_mimage_count = lv_mirror_count(lv);
|
|
*old_log_count = _get_log_count(lv);
|
|
|
|
if (lv->vg->lock_type && !strcmp(lv->vg->lock_type, "sanlock") && lp->keep_mimages) {
|
|
/* FIXME: we need to create a sanlock lock on disk for the new LV. */
|
|
log_error("Unable to split mirrors in VG with lock_type %s", lv->vg->lock_type);
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* Adjusting mimage count?
|
|
*/
|
|
if (!lp->mirrors_supplied && !lp->keep_mimages)
|
|
lp->mirrors = *old_mimage_count;
|
|
else if (lp->mirrors_sign == SIGN_PLUS)
|
|
lp->mirrors = *old_mimage_count + lp->mirrors;
|
|
else if (lp->mirrors_sign == SIGN_MINUS)
|
|
lp->mirrors = (*old_mimage_count > lp->mirrors) ?
|
|
*old_mimage_count - lp->mirrors: 0;
|
|
else
|
|
lp->mirrors += 1;
|
|
|
|
*new_mimage_count = lp->mirrors;
|
|
|
|
/* Too many mimages? */
|
|
if (lp->mirrors > DEFAULT_MIRROR_MAX_IMAGES) {
|
|
log_error("Only up to %d images in mirror supported currently.",
|
|
DEFAULT_MIRROR_MAX_IMAGES);
|
|
return 0;
|
|
}
|
|
|
|
/* Did the user try to subtract more legs than available? */
|
|
if (lp->mirrors < 1) {
|
|
log_error("Unable to reduce images by specified amount - only %d in %s",
|
|
*old_mimage_count, lv->name);
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* FIXME: It would be nice to say what we are adjusting to, but
|
|
* I really don't know whether to specify the # of copies or mimages.
|
|
*/
|
|
if (*old_mimage_count != *new_mimage_count)
|
|
log_verbose("Adjusting mirror image count of %s", lv->name);
|
|
|
|
/* If region size is not given by user - use value from mirror */
|
|
if (lv_is_mirrored(lv) && !lp->region_size_supplied) {
|
|
lp->region_size = first_seg(lv)->region_size;
|
|
log_debug("Copying region size %s from existing mirror.",
|
|
display_size(lv->vg->cmd, lp->region_size));
|
|
}
|
|
|
|
/*
|
|
* Adjust log type
|
|
*
|
|
* If we are converting from a mirror to another mirror or simply
|
|
* changing the log type, we start by assuming they want the log
|
|
* type the same and then parse the given args. OTOH, If we are
|
|
* converting from linear to mirror, then we start from the default
|
|
* position that the user would like a 'disk' log.
|
|
*/
|
|
*new_log_count = (*old_mimage_count > 1) ? *old_log_count : 1;
|
|
if (!lp->corelog && !lp->mirrorlog)
|
|
return 1;
|
|
|
|
*new_log_count = arg_int_value(cmd, mirrorlog_ARG, lp->corelog ? MIRROR_LOG_CORE : DEFAULT_MIRRORLOG);
|
|
|
|
log_verbose("Setting logging type to %s.", get_mirror_log_name(*new_log_count));
|
|
|
|
/*
|
|
* Region size must not change on existing mirrors
|
|
*/
|
|
if (arg_is_set(cmd, regionsize_ARG) && lv_is_mirrored(lv) &&
|
|
(lp->region_size != first_seg(lv)->region_size)) {
|
|
log_error("Mirror log region size cannot be changed on "
|
|
"an existing mirror.");
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* For the most part, we cannot handle multi-segment mirrors. Bail out
|
|
* early if we have encountered one.
|
|
*/
|
|
if (lv_is_mirrored(lv) && dm_list_size(&lv->segments) != 1) {
|
|
log_error("Logical volume %s has multiple "
|
|
"mirror segments.", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
return 1;
|
|
}
|
|
|
|
/*
|
|
* _lvconvert_mirrors_aux
|
|
*
|
|
* Add/remove mirror images and adjust log type. 'operable_pvs'
|
|
* are the set of PVs open to removal or allocation - depending
|
|
* on the operation being performed.
|
|
*/
|
|
static int _lvconvert_mirrors_aux(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct lvconvert_params *lp,
|
|
struct dm_list *operable_pvs,
|
|
uint32_t new_mimage_count,
|
|
uint32_t new_log_count,
|
|
struct dm_list *pvh)
|
|
{
|
|
uint32_t region_size;
|
|
struct lv_segment *seg = first_seg(lv);
|
|
struct logical_volume *layer_lv;
|
|
uint32_t old_mimage_count = lv_mirror_count(lv);
|
|
uint32_t old_log_count = _get_log_count(lv);
|
|
|
|
if ((lp->mirrors == 1) && !lv_is_mirrored(lv)) {
|
|
log_warn("WARNING: Logical volume %s is already not mirrored.",
|
|
display_lvname(lv));
|
|
return 1;
|
|
}
|
|
|
|
if (!(region_size = adjusted_mirror_region_size(cmd, lv->vg->extent_size,
|
|
lv->le_count,
|
|
lp->region_size ? : seg->region_size, 0,
|
|
vg_is_clustered(lv->vg))))
|
|
return_0;
|
|
|
|
if (lv_component_is_active(lv)) {
|
|
log_error("Cannot convert logical volume %s with active component LV(s).",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (!operable_pvs)
|
|
operable_pvs = pvh;
|
|
|
|
/*
|
|
* Up-convert from linear to mirror
|
|
*/
|
|
if (!lv_is_mirrored(lv)) {
|
|
/* FIXME Share code with lvcreate */
|
|
|
|
/*
|
|
* FIXME should we give not only pvh, but also all PVs
|
|
* currently taken by the mirror? Would make more sense from
|
|
* user perspective.
|
|
*/
|
|
if (!lv_add_mirrors(cmd, lv, new_mimage_count - 1, lp->stripes,
|
|
lp->stripe_size, region_size, new_log_count, operable_pvs,
|
|
lp->alloc, MIRROR_BY_LV))
|
|
return_0;
|
|
|
|
if (!arg_is_set(cmd, background_ARG))
|
|
lp->need_polling = 1;
|
|
|
|
goto out;
|
|
}
|
|
|
|
/*
|
|
* Up-convert m-way mirror to n-way mirror
|
|
*/
|
|
if (new_mimage_count > old_mimage_count) {
|
|
if (lv_is_not_synced(lv)) {
|
|
log_error("Can't add mirror to out-of-sync mirrored "
|
|
"LV: use lvchange --resync first.");
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* We allow snapshots of mirrors, but for now, we
|
|
* do not allow up converting mirrors that are under
|
|
* snapshots. The layering logic is somewhat complex,
|
|
* and preliminary test show that the conversion can't
|
|
* seem to get the correct %'age of completion.
|
|
*/
|
|
if (lv_is_origin(lv)) {
|
|
log_error("Can't add additional mirror images to "
|
|
"mirror %s which is under snapshots.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* Is there already a convert in progress? We do not
|
|
* currently allow more than one.
|
|
*/
|
|
if (find_temporary_mirror(lv) || lv_is_converting(lv)) {
|
|
log_error("%s is already being converted. Unable to start another conversion.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* Log addition/removal should be done before the layer
|
|
* insertion to make the end result consistent with
|
|
* linear-to-mirror conversion.
|
|
*/
|
|
if (!_lv_update_log_type(cmd, lp, lv,
|
|
operable_pvs, new_log_count))
|
|
return_0;
|
|
|
|
/* Insert a temporary layer for syncing,
|
|
* only if the original lv is using disk log. */
|
|
if (seg->log_lv && !_insert_lvconvert_layer(cmd, lv)) {
|
|
log_error("Failed to insert resync layer.");
|
|
return 0;
|
|
}
|
|
|
|
/* FIXME: can't have multiple mlogs. force corelog. */
|
|
if (!lv_add_mirrors(cmd, lv,
|
|
new_mimage_count - old_mimage_count,
|
|
lp->stripes, lp->stripe_size,
|
|
region_size, 0U, operable_pvs, lp->alloc,
|
|
MIRROR_BY_LV)) {
|
|
/* FIXME: failure inside library -> move error path processing into library. */
|
|
layer_lv = seg_lv(first_seg(lv), 0);
|
|
if (!remove_layer_from_lv(lv, layer_lv) ||
|
|
(lv_is_active(lv) && !deactivate_lv(cmd, layer_lv)) ||
|
|
!lv_remove(layer_lv) ||
|
|
!vg_write(lv->vg) || !vg_commit(lv->vg)) {
|
|
log_error("ABORTING: Failed to remove "
|
|
"temporary mirror layer %s.",
|
|
display_lvname(layer_lv));
|
|
log_error("Manual cleanup with vgcfgrestore "
|
|
"and dmsetup may be required.");
|
|
return 0;
|
|
}
|
|
|
|
return_0;
|
|
}
|
|
if (seg->log_lv)
|
|
lv->status |= CONVERTING;
|
|
lp->need_polling = 1;
|
|
|
|
goto out_skip_log_convert;
|
|
}
|
|
|
|
/*
|
|
* Down-convert (reduce # of mimages).
|
|
*/
|
|
if (new_mimage_count < old_mimage_count) {
|
|
uint32_t nmc = old_mimage_count - new_mimage_count;
|
|
uint32_t nlc = (!new_log_count || lp->mirrors == 1) ? 1U : 0U;
|
|
|
|
/* FIXME: Why did nlc used to be calculated that way? */
|
|
|
|
/* Reduce number of mirrors */
|
|
if (lp->keep_mimages) {
|
|
if (lp->track_changes) {
|
|
log_error("--trackchanges is not available "
|
|
"to 'mirror' segment type.");
|
|
return 0;
|
|
}
|
|
if (!lv_split_mirror_images(lv, lp->lv_split_name,
|
|
nmc, operable_pvs))
|
|
return_0;
|
|
} else if (!lv_remove_mirrors(cmd, lv, nmc, nlc,
|
|
is_mirror_image_removable, operable_pvs, 0))
|
|
return_0;
|
|
|
|
goto out; /* Just in case someone puts code between */
|
|
}
|
|
|
|
out:
|
|
/*
|
|
* Converting the log type
|
|
*/
|
|
if (lv_is_mirrored(lv) && (old_log_count != new_log_count)) {
|
|
if (!_lv_update_log_type(cmd, lp, lv,
|
|
operable_pvs, new_log_count))
|
|
return_0;
|
|
}
|
|
|
|
out_skip_log_convert:
|
|
|
|
if (!lv_update_and_reload(lv))
|
|
return_0;
|
|
|
|
return 1;
|
|
}
|
|
|
|
int mirror_remove_missing(struct cmd_context *cmd,
|
|
struct logical_volume *lv, int force)
|
|
{
|
|
struct dm_list *failed_pvs;
|
|
int log_count = _get_log_count(lv) - _failed_logs_count(lv);
|
|
|
|
if (!(failed_pvs = _failed_pv_list(lv->vg)))
|
|
return_0;
|
|
|
|
if (force && _failed_mirrors_count(lv) == (int)lv_mirror_count(lv)) {
|
|
log_error("No usable images left in %s.", display_lvname(lv));
|
|
return lv_remove_with_dependencies(cmd, lv, DONT_PROMPT, 0);
|
|
}
|
|
|
|
/*
|
|
* We must adjust the log first, or the entire mirror
|
|
* will get stuck during a suspend.
|
|
*/
|
|
if (!_lv_update_mirrored_log(lv, failed_pvs, log_count))
|
|
return_0;
|
|
|
|
if (_failed_mirrors_count(lv) > 0 &&
|
|
!lv_remove_mirrors(cmd, lv, _failed_mirrors_count(lv),
|
|
log_count ? 0U : 1U,
|
|
_is_partial_lv, NULL, 0))
|
|
return_0;
|
|
|
|
if (lv_is_mirrored(lv) &&
|
|
!_lv_update_log_type(cmd, NULL, lv, failed_pvs, log_count))
|
|
return_0;
|
|
|
|
if (!lv_update_and_reload(lv))
|
|
return_0;
|
|
|
|
return 1;
|
|
}
|
|
|
|
/*
|
|
* _lvconvert_mirrors_repair
|
|
*
|
|
* This function operates in two phases. First, all of the bad
|
|
* devices are removed from the mirror. Then, if desired by the
|
|
* user, the devices are replaced.
|
|
*
|
|
* 'old_mimage_count' and 'old_log_count' are there so we know
|
|
* what to convert to after the removal of devices.
|
|
*/
|
|
static int _lvconvert_mirrors_repair(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct lvconvert_params *lp,
|
|
struct dm_list *pvh)
|
|
{
|
|
int failed_logs;
|
|
int failed_mimages;
|
|
int replace_logs = 0;
|
|
int replace_mimages = 0;
|
|
uint32_t log_count;
|
|
|
|
uint32_t original_mimages = lv_mirror_count(lv);
|
|
uint32_t original_logs = _get_log_count(lv);
|
|
|
|
cmd->partial_activation = 1;
|
|
lp->need_polling = 0;
|
|
|
|
lv_check_transient(lv); /* TODO check this in lib for all commands? */
|
|
|
|
if (!lv_is_partial(lv)) {
|
|
log_print_unless_silent("Volume %s is consistent. Nothing to repair.",
|
|
display_lvname(lv));
|
|
return 1;
|
|
}
|
|
|
|
failed_mimages = _failed_mirrors_count(lv);
|
|
failed_logs = _failed_logs_count(lv);
|
|
|
|
/* Retain existing region size in case we need it later */
|
|
if (!lp->region_size)
|
|
lp->region_size = first_seg(lv)->region_size;
|
|
|
|
if (!mirror_remove_missing(cmd, lv, 0))
|
|
return_0;
|
|
|
|
if (failed_mimages)
|
|
log_print_unless_silent("Mirror status: %d of %d images failed.",
|
|
failed_mimages, original_mimages);
|
|
|
|
/*
|
|
* Count the failed log devices
|
|
*/
|
|
if (failed_logs)
|
|
log_print_unless_silent("Mirror log status: %d of %d images failed.",
|
|
failed_logs, original_logs);
|
|
|
|
/*
|
|
* Find out our policies
|
|
*/
|
|
_lvconvert_mirrors_repair_ask(cmd, failed_logs, failed_mimages,
|
|
&replace_logs, &replace_mimages);
|
|
|
|
/*
|
|
* Second phase - replace faulty devices
|
|
*/
|
|
lp->mirrors = replace_mimages ? original_mimages : (original_mimages - failed_mimages);
|
|
|
|
/*
|
|
* It does not make sense to replace the log if the volume is no longer
|
|
* a mirror.
|
|
*/
|
|
if (lp->mirrors == 1)
|
|
replace_logs = 0;
|
|
|
|
log_count = replace_logs ? original_logs : (original_logs - failed_logs);
|
|
|
|
while (replace_mimages || replace_logs) {
|
|
log_warn("WARNING: Trying to up-convert to %d images, %d logs.", lp->mirrors, log_count);
|
|
if (_lvconvert_mirrors_aux(cmd, lv, lp, NULL,
|
|
lp->mirrors, log_count, pvh))
|
|
break;
|
|
if (lp->mirrors > 2)
|
|
--lp->mirrors;
|
|
else if (log_count > 0)
|
|
--log_count;
|
|
else
|
|
break; /* nowhere to go, anymore... */
|
|
}
|
|
|
|
if (replace_mimages && lv_mirror_count(lv) != original_mimages)
|
|
log_warn("WARNING: Failed to replace %d of %d images in volume %s.",
|
|
original_mimages - lv_mirror_count(lv), original_mimages,
|
|
display_lvname(lv));
|
|
if (replace_logs && _get_log_count(lv) != original_logs)
|
|
log_warn("WARNING: Failed to replace %d of %d logs in volume %s.",
|
|
original_logs - _get_log_count(lv), original_logs,
|
|
display_lvname(lv));
|
|
|
|
/* if (!arg_is_set(cmd, use_policies_ARG) && (lp->mirrors != old_mimage_count
|
|
|| log_count != old_log_count))
|
|
return 0; */
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_validate_thin(struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
if (!lv_is_thin_pool(lv) && !lv_is_thin_volume(lv))
|
|
return 1;
|
|
|
|
log_error("Converting thin%s segment type for %s to %s is not supported.",
|
|
lv_is_thin_pool(lv) ? " pool" : "",
|
|
display_lvname(lv), lp->segtype->name);
|
|
|
|
if (lv_is_thin_volume(lv))
|
|
return 0;
|
|
|
|
/* Give advice for thin pool conversion */
|
|
log_error("For pool data volume conversion use %s.",
|
|
display_lvname(seg_lv(first_seg(lv), 0)));
|
|
log_error("For pool metadata volume conversion use %s.",
|
|
display_lvname(first_seg(lv)->metadata_lv));
|
|
|
|
return 0;
|
|
}
|
|
|
|
/* Check for raid1 split trackchanges image to reject conversions on it. */
|
|
static int _raid_split_image_conversion(struct logical_volume *lv)
|
|
{
|
|
const char *s;
|
|
char raidlv_name[NAME_LEN];
|
|
const struct logical_volume *tmp_lv;
|
|
|
|
if (lv_is_raid_with_tracking(lv)) {
|
|
log_error("Conversion of tracking raid1 LV %s is not supported.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_raid_image(lv) &&
|
|
(s = strstr(lv->name, "_rimage_"))) {
|
|
dm_strncpy(raidlv_name, lv->name, s - lv->name);
|
|
|
|
if (!(tmp_lv = find_lv(lv->vg, raidlv_name))) {
|
|
log_error("Failed to find RaidLV of RAID subvolume %s.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_raid_with_tracking(tmp_lv)) {
|
|
log_error("Conversion of tracked raid1 subvolume %s is not supported.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
return 1;
|
|
}
|
|
|
|
/*
|
|
* _lvconvert_mirrors
|
|
*
|
|
* Determine what is being done. Are we doing a conversion, repair, or
|
|
* collapsing a stack? Once determined, call helper functions.
|
|
*/
|
|
static int _lvconvert_mirrors(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
uint32_t old_mimage_count = 0;
|
|
uint32_t old_log_count = 0;
|
|
uint32_t new_mimage_count = 0;
|
|
uint32_t new_log_count = 0;
|
|
|
|
if (!_raid_split_image_conversion(lv))
|
|
return_0;
|
|
|
|
if ((lp->corelog || lp->mirrorlog) && *lp->type_str && strcmp(lp->type_str, SEG_TYPE_NAME_MIRROR)) {
|
|
log_error("--corelog and --mirrorlog are only compatible with mirror devices.");
|
|
return 0;
|
|
}
|
|
|
|
if (!_lvconvert_validate_thin(lv, lp))
|
|
return_0;
|
|
|
|
if (lv_is_thin_type(lv)) {
|
|
log_error("Mirror segment type cannot be used for thinpool%s.\n"
|
|
"Try \"%s\" segment type instead.",
|
|
lv_is_thin_pool_data(lv) ? "s" : " metadata",
|
|
SEG_TYPE_NAME_RAID1);
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_cache_type(lv)) {
|
|
log_error("Mirrors are not yet supported on cache LVs %s.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (_linear_type_requested(lp->type_str)) {
|
|
if (arg_is_set(cmd, mirrors_ARG) && (arg_uint_value(cmd, mirrors_ARG, 0) != 0)) {
|
|
log_error("Cannot specify mirrors with linear type.");
|
|
return 0;
|
|
}
|
|
lp->mirrors_supplied = 1;
|
|
lp->mirrors = 0;
|
|
}
|
|
|
|
/* Adjust mimage and/or log count */
|
|
if (!_lvconvert_mirrors_parse_params(cmd, lv, lp,
|
|
&old_mimage_count, &old_log_count,
|
|
&new_mimage_count, &new_log_count))
|
|
return_0;
|
|
|
|
if (((old_mimage_count < new_mimage_count && old_log_count > new_log_count) ||
|
|
(old_mimage_count > new_mimage_count && old_log_count < new_log_count)) &&
|
|
lp->pv_count) {
|
|
log_error("Cannot both allocate and free extents when "
|
|
"specifying physical volumes to use.");
|
|
log_error("Please specify the operation in two steps.");
|
|
return 0;
|
|
}
|
|
|
|
/* Nothing to do? (Probably finishing collapse.) */
|
|
if ((old_mimage_count == new_mimage_count) &&
|
|
(old_log_count == new_log_count))
|
|
return 1;
|
|
|
|
if (!_lvconvert_mirrors_aux(cmd, lv, lp, NULL,
|
|
new_mimage_count, new_log_count, lp->pvh))
|
|
return_0;
|
|
|
|
if (!lp->need_polling)
|
|
log_print_unless_silent("Logical volume %s converted.",
|
|
display_lvname(lv));
|
|
else
|
|
log_print_unless_silent("Logical volume %s being converted.",
|
|
display_lvname(lv));
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _is_valid_raid_conversion(const struct segment_type *from_segtype,
|
|
const struct segment_type *to_segtype)
|
|
{
|
|
if (!from_segtype)
|
|
return 1;
|
|
|
|
/* linear/striped/raid0 <-> striped/raid0/linear (restriping via raid) */
|
|
if (segtype_is_striped(from_segtype) && segtype_is_striped(to_segtype))
|
|
return 0;
|
|
|
|
if (from_segtype == to_segtype)
|
|
return 1;
|
|
|
|
if (!segtype_is_raid(from_segtype) && !segtype_is_raid(to_segtype))
|
|
return_0; /* Not converting to or from RAID? */
|
|
|
|
return 1;
|
|
}
|
|
|
|
/* Check for dm-raid target supporting raid4 conversion properly. */
|
|
static int _raid4_conversion_supported(struct logical_volume *lv, struct lvconvert_params *lp)
|
|
{
|
|
int ret = 1;
|
|
struct lv_segment *seg = first_seg(lv);
|
|
|
|
if (seg_is_raid4(seg))
|
|
ret = raid4_is_supported(lv->vg->cmd, seg->segtype);
|
|
else if (segtype_is_raid4(lp->segtype))
|
|
ret = raid4_is_supported(lv->vg->cmd, lp->segtype);
|
|
|
|
if (ret)
|
|
return 1;
|
|
|
|
log_error("Cannot convert %s LV %s to %s.",
|
|
lvseg_name(seg), display_lvname(lv), lp->segtype->name);
|
|
return 0;
|
|
}
|
|
|
|
static int _lvconvert_raid(struct logical_volume *lv, struct lvconvert_params *lp)
|
|
{
|
|
int image_count = 0;
|
|
int images_reduced = 0;
|
|
int type_enforced = 0;
|
|
struct cmd_context *cmd = lv->vg->cmd;
|
|
struct lv_segment *seg = first_seg(lv);
|
|
|
|
if (!_raid_split_image_conversion(lv))
|
|
return_0;
|
|
|
|
if (_linear_type_requested(lp->type_str)) {
|
|
if (arg_is_set(cmd, mirrors_ARG) && (arg_uint_value(cmd, mirrors_ARG, 0) != 0)) {
|
|
log_error("Cannot specify mirrors with linear type.");
|
|
return 0;
|
|
}
|
|
lp->mirrors_supplied = 1;
|
|
lp->mirrors = 0;
|
|
}
|
|
|
|
if (!_lvconvert_validate_thin(lv, lp))
|
|
return_0;
|
|
|
|
if (!_is_valid_raid_conversion(seg->segtype, lp->segtype) &&
|
|
!lp->mirrors_supplied)
|
|
goto try_new_takeover_or_reshape;
|
|
|
|
if (seg_is_striped(seg) && !lp->mirrors_supplied)
|
|
goto try_new_takeover_or_reshape;
|
|
|
|
if (seg_is_linear(seg) && !lp->mirrors_supplied)
|
|
goto try_new_takeover_or_reshape;
|
|
|
|
/* Change number of RAID1 images */
|
|
if (lp->mirrors_supplied || lp->keep_mimages) {
|
|
image_count = lv_raid_image_count(lv);
|
|
if (lp->mirrors_sign == SIGN_PLUS)
|
|
image_count += lp->mirrors;
|
|
else if (lp->mirrors_sign == SIGN_MINUS)
|
|
image_count -= lp->mirrors;
|
|
else
|
|
image_count = lp->mirrors + 1;
|
|
|
|
images_reduced = (image_count < (int) lv_raid_image_count(lv));
|
|
|
|
if (image_count < 1) {
|
|
log_error("Unable to %s images by specified amount.",
|
|
lp->keep_mimages ? "split" : "reduce");
|
|
return 0;
|
|
}
|
|
|
|
/* --trackchanges requires --splitmirrors which always has SIGN_MINUS */
|
|
if (lp->track_changes && lp->mirrors != 1) {
|
|
log_error("Exactly one image must be split off from %s when tracking changes.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (!*lp->type_str) {
|
|
lp->type_str = SEG_TYPE_NAME_RAID1;
|
|
if (!(lp->segtype = get_segtype_from_string(lv->vg->cmd, SEG_TYPE_NAME_RAID1)))
|
|
return_0;
|
|
type_enforced = 1;
|
|
}
|
|
}
|
|
|
|
if ((lp->corelog || lp->mirrorlog) && strcmp(lp->type_str, SEG_TYPE_NAME_MIRROR)) {
|
|
log_error("--corelog and --mirrorlog are only compatible with mirror devices");
|
|
return 0;
|
|
}
|
|
|
|
if (lp->track_changes)
|
|
return lv_raid_split_and_track(lv, lp->yes, lp->pvh);
|
|
|
|
if (lp->keep_mimages)
|
|
return lv_raid_split(lv, lp->yes, lp->lv_split_name, image_count, lp->pvh);
|
|
|
|
if (lp->mirrors_supplied) {
|
|
if (seg_is_linear(seg) || seg_is_raid1(seg)) { /* ??? */
|
|
if (!*lp->type_str || !strcmp(lp->type_str, SEG_TYPE_NAME_RAID1) || !strcmp(lp->type_str, SEG_TYPE_NAME_LINEAR) ||
|
|
(!strcmp(lp->type_str, SEG_TYPE_NAME_STRIPED) && image_count == 1)) {
|
|
if (image_count > DEFAULT_RAID1_MAX_IMAGES) {
|
|
log_error("Only up to %u mirrors in %s LV %s supported currently.",
|
|
DEFAULT_RAID1_MAX_IMAGES, lp->segtype->name, display_lvname(lv));
|
|
return 0;
|
|
}
|
|
if (!seg_is_raid1(seg) && lv_raid_has_integrity(lv)) {
|
|
log_error("Cannot add raid images with integrity for this raid level.");
|
|
return 0;
|
|
}
|
|
if (!lv_raid_change_image_count(lv, lp->yes, image_count,
|
|
(lp->region_size_supplied || !seg->region_size) ?
|
|
lp->region_size : seg->region_size , lp->pvh))
|
|
return_0;
|
|
|
|
if (lv_raid_has_integrity(lv) && !images_reduced) {
|
|
struct integrity_settings *isettings = NULL;
|
|
if (!lv_get_raid_integrity_settings(lv, &isettings))
|
|
return_0;
|
|
if (!lv_add_integrity_to_raid(lv, isettings, lp->pvh, NULL))
|
|
return_0;
|
|
}
|
|
|
|
log_print_unless_silent("Logical volume %s successfully converted.",
|
|
display_lvname(lv));
|
|
|
|
return 1;
|
|
}
|
|
}
|
|
goto try_new_takeover_or_reshape;
|
|
}
|
|
|
|
if ((seg_is_linear(seg) || seg_is_striped(seg) || seg_is_mirrored(seg) || lv_is_raid(lv)) &&
|
|
(lp->type_str && lp->type_str[0])) {
|
|
/* Activation is required later which precludes existing supported raid0 segment */
|
|
if ((seg_is_any_raid0(seg) || segtype_is_any_raid0(lp->segtype)) &&
|
|
!(lp->target_attr & RAID_FEATURE_RAID0)) {
|
|
log_error("RAID module does not support RAID0.");
|
|
return 0;
|
|
}
|
|
|
|
/* Activation is required later which precludes existing supported raid4 segment */
|
|
if (!_raid4_conversion_supported(lv, lp))
|
|
return_0;
|
|
|
|
/* Activation is required later which precludes existing supported raid10 segment */
|
|
if ((seg_is_raid10(seg) || segtype_is_raid10(lp->segtype)) &&
|
|
!(lp->target_attr & RAID_FEATURE_RAID10)) {
|
|
log_error("RAID module does not support RAID10.");
|
|
return 0;
|
|
}
|
|
|
|
if (lv_raid_has_integrity(lv)) {
|
|
/* FIXME: which conversions are happening here? */
|
|
log_error("This conversion is not supported for raid with integrity.");
|
|
return 0;
|
|
}
|
|
|
|
/* FIXME This needs changing globally. */
|
|
if (!arg_is_set(cmd, stripes_long_ARG))
|
|
lp->stripes = 0;
|
|
if (!type_enforced && !arg_is_set(cmd, type_ARG))
|
|
lp->segtype = NULL;
|
|
if (!arg_is_set(cmd, regionsize_ARG))
|
|
lp->region_size = 0;
|
|
|
|
if (!lv_raid_convert(lv, lp->segtype,
|
|
lp->yes, lp->force, lp->stripes, lp->stripe_size_supplied, lp->stripe_size,
|
|
lp->region_size, lp->pvh))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Logical volume %s successfully converted.",
|
|
display_lvname(lv));
|
|
return 1;
|
|
}
|
|
|
|
try_new_takeover_or_reshape:
|
|
if (lv_raid_has_integrity(lv)) {
|
|
/* FIXME: which conversions are happening here? */
|
|
log_error("This conversion is not supported for raid with integrity.");
|
|
return 0;
|
|
}
|
|
|
|
if (!_raid4_conversion_supported(lv, lp))
|
|
return 0;
|
|
|
|
/* FIXME This needs changing globally. */
|
|
if (!arg_is_set(cmd, stripes_long_ARG))
|
|
lp->stripes = 0;
|
|
if (!type_enforced && !arg_is_set(cmd, type_ARG))
|
|
lp->segtype = NULL;
|
|
|
|
if (!lv_raid_convert(lv, lp->segtype,
|
|
lp->yes, lp->force, lp->stripes, lp->stripe_size_supplied, lp->stripe_size,
|
|
(lp->region_size_supplied || !seg->region_size) ?
|
|
lp->region_size : seg->region_size , lp->pvh))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Logical volume %s successfully converted.",
|
|
display_lvname(lv));
|
|
return 1;
|
|
}
|
|
|
|
/*
|
|
* Functions called to perform a specific operation on a specific LV type.
|
|
*
|
|
* _convert_<lvtype>_<operation>
|
|
*
|
|
* For cases where an operation does not apply to the LV itself, but
|
|
* is implicitly redirected to a sub-LV, these functions locate the
|
|
* correct sub-LV and call the operation on that sub-LV. If a sub-LV
|
|
* of the proper type is not found, these functions report the error.
|
|
*
|
|
* FIXME: the _lvconvert_foo() functions can be cleaned up since they
|
|
* are now only called for valid combinations of LV type and operation.
|
|
* After that happens, the code remaining in those functions can be
|
|
* moved into the _convert_lvtype_operation() functions below.
|
|
*/
|
|
|
|
/*
|
|
* Change the number of images in a mirror LV.
|
|
* lvconvert --mirrors Number LV
|
|
*/
|
|
static int _convert_mirror_number(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_mirrors(cmd, lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Split images from a mirror LV and use them to create a new LV.
|
|
* lvconvert --splitmirrors Number LV
|
|
*
|
|
* Required options:
|
|
* --name Name
|
|
*/
|
|
|
|
static int _convert_mirror_splitmirrors(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_mirrors(cmd, lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Change the type of log used by a mirror LV.
|
|
* lvconvert --mirrorlog Type LV
|
|
*/
|
|
static int _convert_mirror_log(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_mirrors(cmd, lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Convert mirror LV to linear LV.
|
|
* lvconvert --type linear LV
|
|
*
|
|
* Alternate syntax:
|
|
* lvconvert --mirrors 0 LV
|
|
*/
|
|
static int _convert_mirror_linear(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_mirrors(cmd, lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Convert mirror LV to raid1 LV.
|
|
* lvconvert --type raid1 LV
|
|
*/
|
|
static int _convert_mirror_raid(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_raid(lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Change the number of images in a raid1 LV.
|
|
* lvconvert --mirrors Number LV
|
|
*/
|
|
static int _convert_raid_number(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_raid(lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Split images from a raid1 LV and use them to create a new LV.
|
|
* lvconvert --splitmirrors Number LV
|
|
*
|
|
* Required options:
|
|
* --trackchanges | --name Name
|
|
*/
|
|
static int _convert_raid_splitmirrors(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
/* FIXME: split the splitmirrors section out of _lvconvert_raid and call it here. */
|
|
return _lvconvert_raid(lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Convert a raid* LV to use a different raid level.
|
|
* lvconvert --type raid* LV
|
|
*/
|
|
static int _convert_raid_raid(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_raid(lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Convert a raid* LV to a mirror LV.
|
|
* lvconvert --type mirror LV
|
|
*/
|
|
static int _convert_raid_mirror(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_raid(lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Convert a raid* LV to a striped LV.
|
|
* lvconvert --type striped LV
|
|
*/
|
|
static int _convert_raid_striped(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_raid(lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Convert a raid* LV to a linear LV.
|
|
* lvconvert --type linear LV
|
|
*/
|
|
static int _convert_raid_linear(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_raid(lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Convert a striped/linear LV to a mirror LV.
|
|
* lvconvert --type mirror LV
|
|
*
|
|
* Required options:
|
|
* --mirrors Number
|
|
*
|
|
* Alternate syntax:
|
|
* This is equivalent to above when global/mirror_segtype_default="mirror".
|
|
* lvconvert --mirrors Number LV
|
|
*/
|
|
static int _convert_striped_mirror(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_mirrors(cmd, lv, lp);
|
|
}
|
|
|
|
/*
|
|
* Convert a striped/linear LV to a raid* LV.
|
|
* lvconvert --type raid* LV
|
|
*
|
|
* Required options:
|
|
* --mirrors Number
|
|
*
|
|
* Alternate syntax:
|
|
* This is equivalent to above when global/mirror_segtype_default="raid1".
|
|
* lvconvert --mirrors Number LV
|
|
*/
|
|
static int _convert_striped_raid(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
return _lvconvert_raid(lv, lp);
|
|
}
|
|
|
|
static int _convert_mirror(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
if (arg_is_set(cmd, mirrors_ARG))
|
|
return _convert_mirror_number(cmd, lv, lp);
|
|
|
|
if (arg_is_set(cmd, splitmirrors_ARG))
|
|
return _convert_mirror_splitmirrors(cmd, lv, lp);
|
|
|
|
if (arg_is_set(cmd, mirrorlog_ARG) || arg_is_set(cmd, corelog_ARG))
|
|
return _convert_mirror_log(cmd, lv, lp);
|
|
|
|
if (_linear_type_requested(lp->type_str))
|
|
return _convert_mirror_linear(cmd, lv, lp);
|
|
|
|
if (segtype_is_raid(lp->segtype))
|
|
return _convert_mirror_raid(cmd, lv, lp);
|
|
|
|
log_error("Unknown operation on mirror LV %s.", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
static int _convert_raid(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
if (arg_is_set(cmd, mirrors_ARG))
|
|
return _convert_raid_number(cmd, lv, lp);
|
|
|
|
if (arg_is_set(cmd, splitmirrors_ARG))
|
|
return _convert_raid_splitmirrors(cmd, lv, lp);
|
|
|
|
if (segtype_is_raid(lp->segtype))
|
|
return _convert_raid_raid(cmd, lv, lp);
|
|
|
|
if (segtype_is_mirror(lp->segtype))
|
|
return _convert_raid_mirror(cmd, lv, lp);
|
|
|
|
if (!strcmp(lp->type_str, SEG_TYPE_NAME_STRIPED))
|
|
return _convert_raid_striped(cmd, lv, lp);
|
|
|
|
if (_linear_type_requested(lp->type_str))
|
|
return _convert_raid_linear(cmd, lv, lp);
|
|
|
|
log_error("Unknown operation on raid LV %s.", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
static int _convert_striped(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
const char *mirrors_type = find_config_tree_str(cmd, global_mirror_segtype_default_CFG, NULL);
|
|
int raid_type = *lp->type_str && !strncmp(lp->type_str, "raid", 4);
|
|
|
|
if (!raid_type) {
|
|
if (!strcmp(lp->type_str, SEG_TYPE_NAME_MIRROR))
|
|
return _convert_striped_mirror(cmd, lv, lp);
|
|
|
|
/* --mirrors can mean --type mirror or --type raid1 depending on config setting. */
|
|
|
|
if (arg_is_set(cmd, mirrors_ARG) && mirrors_type && !strcmp(mirrors_type, SEG_TYPE_NAME_MIRROR))
|
|
return _convert_striped_mirror(cmd, lv, lp);
|
|
}
|
|
|
|
if (arg_is_set(cmd, mirrors_ARG) && mirrors_type && !strcmp(mirrors_type, SEG_TYPE_NAME_RAID1))
|
|
return _convert_striped_raid(cmd, lv, lp);
|
|
|
|
if (segtype_is_striped(lp->segtype) || segtype_is_raid(lp->segtype))
|
|
return _convert_striped_raid(cmd, lv, lp);
|
|
|
|
log_error("Unknown operation on striped or linear LV %s.", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
static int _lvconvert_raid_types(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct lvconvert_params *lp)
|
|
{
|
|
struct lv_segment *seg = first_seg(lv);
|
|
int ret = 0;
|
|
|
|
/* If LV is inactive here, ensure it's not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
return_ECMD_FAILED;
|
|
|
|
/* Set up segtype either from type_str or else to match the existing one. */
|
|
if (!*lp->type_str)
|
|
lp->segtype = seg->segtype;
|
|
else if (!(lp->segtype = get_segtype_from_string(cmd, lp->type_str)))
|
|
goto_out;
|
|
|
|
if (!strcmp(lp->type_str, SEG_TYPE_NAME_MIRROR)) {
|
|
if (!lp->mirrors_supplied && !seg_is_raid1(seg)) {
|
|
log_error("Conversions to --type mirror require -m/--mirrors");
|
|
goto out;
|
|
}
|
|
}
|
|
|
|
/* lv->segtype can't be NULL */
|
|
if (activation() && lp->segtype->ops->target_present &&
|
|
!lp->segtype->ops->target_present(cmd, NULL, &lp->target_attr)) {
|
|
log_error("%s: Required device-mapper target(s) not "
|
|
"detected in your kernel.", lp->segtype->name);
|
|
goto out;
|
|
}
|
|
|
|
/* Process striping parameters */
|
|
/* FIXME This is incomplete */
|
|
if (_mirror_or_raid_type_requested(cmd, lp->type_str) || _raid0_type_requested(lp->type_str) ||
|
|
_striped_type_requested(lp->type_str) || lp->mirrorlog || lp->corelog) {
|
|
if (!arg_is_set(cmd, type_ARG))
|
|
lp->segtype = first_seg(lv)->segtype;
|
|
/* FIXME Handle +/- adjustments too? */
|
|
if (!get_stripe_params(cmd, lp->segtype, &lp->stripes, &lp->stripe_size, &lp->stripes_supplied, &lp->stripe_size_supplied))
|
|
goto_out;
|
|
|
|
if (_raid0_type_requested(lp->type_str) || _striped_type_requested(lp->type_str))
|
|
/* FIXME Shouldn't need to override get_stripe_params which defaults to 1 stripe (i.e. linear)! */
|
|
/* The default keeps existing number of stripes, handled inside the library code */
|
|
if (!arg_is_set(cmd, stripes_long_ARG))
|
|
lp->stripes = 0;
|
|
}
|
|
|
|
if (lv_is_cache(lv))
|
|
lv = seg_lv(first_seg(lv), 0);
|
|
|
|
if (lv_is_vdo_pool(lv))
|
|
return _lvconvert_raid_types(cmd, seg_lv(first_seg(lv), 0), lp);
|
|
|
|
if (lv_is_mirror(lv)) {
|
|
ret = _convert_mirror(cmd, lv, lp);
|
|
goto out;
|
|
}
|
|
|
|
if (lv_is_raid(lv)) {
|
|
ret = _convert_raid(cmd, lv, lp);
|
|
goto out;
|
|
}
|
|
|
|
/*
|
|
* FIXME: add lv_is_striped() and lv_is_linear()?
|
|
* This does not include raid0 which is caught by the test above.
|
|
* If operations differ between striped and linear, split this case.
|
|
*/
|
|
if (segtype_is_striped(seg->segtype) || segtype_is_linear(seg->segtype)) {
|
|
ret = _convert_striped(cmd, lv, lp);
|
|
goto out;
|
|
}
|
|
|
|
/*
|
|
* The intention is to explicitly check all cases above and never
|
|
* reach here, but this covers anything that was missed.
|
|
*/
|
|
log_error("Cannot convert LV %s.", display_lvname(lv));
|
|
|
|
out:
|
|
return ret ? ECMD_PROCESSED : ECMD_FAILED;
|
|
}
|
|
|
|
static int _lvconvert_splitsnapshot(struct cmd_context *cmd, struct logical_volume *cow)
|
|
{
|
|
struct volume_group *vg = cow->vg;
|
|
const char *cow_name = display_lvname(cow);
|
|
|
|
if (!lv_is_cow(cow)) {
|
|
log_error(INTERNAL_ERROR "Volume %s is not a COW.", cow_name);
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_virtual_origin(origin_from_cow(cow))) {
|
|
log_error("Unable to split off snapshot %s with virtual origin.", cow_name);
|
|
return 0;
|
|
}
|
|
|
|
if (vg_is_shared(vg)) {
|
|
/* FIXME: we need to create a lock for the new LV. */
|
|
log_error("Unable to split snapshots in VG with lock_type %s.", vg->lock_type);
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_active(cow)) {
|
|
if (!lv_check_not_in_use(cow, 1))
|
|
return_0;
|
|
|
|
if ((arg_count(cmd, force_ARG) == PROMPT) &&
|
|
!arg_count(cmd, yes_ARG) &&
|
|
lv_is_visible(cow) &&
|
|
lv_is_active(cow)) {
|
|
if (yes_no_prompt("Do you really want to split off active "
|
|
"logical volume %s? [y/n]: ", display_lvname(cow)) == 'n') {
|
|
log_error("Logical volume %s not split.", display_lvname(cow));
|
|
return 0;
|
|
}
|
|
}
|
|
}
|
|
|
|
log_verbose("Splitting snapshot %s from its origin.", display_lvname(cow));
|
|
|
|
if (!vg_remove_snapshot(cow))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Logical Volume %s split from its origin.", display_lvname(cow));
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_split_and_keep_cachevol(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *lv_fast)
|
|
{
|
|
char cvol_name[NAME_LEN];
|
|
struct lv_segment *cache_seg = first_seg(lv);
|
|
int cache_mode = cache_seg->cache_mode;
|
|
int direct_detach = 0;
|
|
|
|
if (!archive(lv->vg))
|
|
return_0;
|
|
|
|
log_debug("Detaching cachevol %s from LV %s.", display_lvname(lv_fast), display_lvname(lv));
|
|
|
|
/*
|
|
* Allow forcible detach without activating or flushing
|
|
* in case the cache is corrupt/damaged/invalid.
|
|
* This would generally be done to rescue data from
|
|
* the origin if the cache could not be repaired.
|
|
*/
|
|
if (!lv_is_active(lv) && arg_count(cmd, force_ARG))
|
|
direct_detach = 1;
|
|
|
|
/*
|
|
* Detaching a writeback cache generally requires flushing;
|
|
* doing otherwise can mean data loss/corruption.
|
|
* If the cache devices are missing, the cache can't be
|
|
* flushed, so require the user to use a force option to
|
|
* detach the cache in this case.
|
|
*/
|
|
if ((cache_mode != CACHE_MODE_WRITETHROUGH) && lv_is_partial(lv_fast)) {
|
|
if (!arg_count(cmd, force_ARG)) {
|
|
log_warn("WARNING: writeback cache on %s is not complete and cannot be flushed.", display_lvname(lv_fast));
|
|
log_warn("WARNING: cannot detach writeback cache from %s without --force.", display_lvname(lv));
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
direct_detach = 1;
|
|
}
|
|
|
|
if (direct_detach) {
|
|
log_warn("WARNING: Data may be lost by detaching writeback cache without flushing.");
|
|
|
|
if (!arg_count(cmd, yes_ARG) &&
|
|
yes_no_prompt("Detach writeback cache %s from %s without flushing data?",
|
|
display_lvname(lv_fast), display_lvname(lv)) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
|
|
/* Switch internally to WRITETHROUGH which does not require flushing */
|
|
cache_seg->cache_mode = CACHE_MODE_WRITETHROUGH;
|
|
}
|
|
|
|
if (!lv_cache_remove(lv))
|
|
return_0;
|
|
|
|
/* Cut off suffix _cvol */
|
|
if (!drop_lvname_suffix(cvol_name, lv_fast->name, "cvol")) {
|
|
/* likely older instance of metadata */
|
|
log_debug("LV %s has no suffix for cachevol (skipping rename).",
|
|
display_lvname(lv_fast));
|
|
} else if (!lv_uniq_rename_update(cmd, lv_fast, cvol_name, 0))
|
|
return_0;
|
|
|
|
if (!vg_write(lv->vg) || !vg_commit(lv->vg))
|
|
return_0;
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_split_and_remove_cachevol(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *lv_fast)
|
|
{
|
|
if (!_lvconvert_split_and_keep_cachevol(cmd, lv, lv_fast))
|
|
return_0;
|
|
|
|
if (lvremove_single(cmd, lv_fast, NULL) != ECMD_PROCESSED)
|
|
return_0;
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_split_and_keep_cachepool(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *lv_fast)
|
|
{
|
|
char name[NAME_LEN];
|
|
|
|
if (!archive(lv->vg))
|
|
return_0;
|
|
|
|
log_debug("Detaching cachepool %s from LV %s.", display_lvname(lv_fast), display_lvname(lv));
|
|
|
|
if (vg_missing_pv_count(lv->vg)) {
|
|
log_error("Cannot split cache pool while PVs are missing, see --uncache to delete cache pool.");
|
|
return 0;
|
|
}
|
|
|
|
if (!lv_cache_remove(lv))
|
|
return_0;
|
|
|
|
/* Cut off suffix _cpool */
|
|
if (!drop_lvname_suffix(name, lv_fast->name, "cpool")) {
|
|
/* likely older instance of metadata */
|
|
log_debug("LV %s has no suffix for cachepool (skipping rename).",
|
|
display_lvname(lv_fast));
|
|
} else if (!lv_uniq_rename_update(cmd, lv_fast, name, 0))
|
|
return_0;
|
|
|
|
if (!vg_write(lv->vg) || !vg_commit(lv->vg))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Logical volume %s is not cached and %s is unused.",
|
|
display_lvname(lv), display_lvname(lv_fast));
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_split_and_remove_cachepool(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *cachepool_lv)
|
|
{
|
|
struct lv_segment *seg;
|
|
struct logical_volume *remove_lv;
|
|
|
|
seg = first_seg(lv);
|
|
|
|
if (lv_is_partial(seg_lv(seg, 0))) {
|
|
log_warn("WARNING: Cache origin logical volume %s is missing.",
|
|
display_lvname(seg_lv(seg, 0)));
|
|
remove_lv = lv; /* When origin is missing, drop everything */
|
|
} else
|
|
remove_lv = seg->pool_lv;
|
|
|
|
if (lv_is_partial(seg_lv(first_seg(seg->pool_lv), 0)))
|
|
log_warn("WARNING: Cache pool data logical volume %s is missing.",
|
|
display_lvname(seg_lv(first_seg(seg->pool_lv), 0)));
|
|
|
|
if (lv_is_partial(first_seg(seg->pool_lv)->metadata_lv))
|
|
log_warn("WARNING: Cache pool metadata logical volume %s is missing.",
|
|
display_lvname(first_seg(seg->pool_lv)->metadata_lv));
|
|
|
|
/* TODO: Check for failed cache as well to get prompting? */
|
|
if (lv_is_partial(lv)) {
|
|
if (first_seg(seg->pool_lv)->cache_mode != CACHE_MODE_WRITETHROUGH) {
|
|
if (!arg_count(cmd, force_ARG)) {
|
|
log_error("Conversion aborted.");
|
|
log_error("Cannot uncache writeback cache volume %s without --force.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
log_warn("WARNING: Uncaching of partially missing %s cache volume %s might destroy your data.",
|
|
cache_mode_num_to_str(first_seg(seg->pool_lv)->cache_mode), display_lvname(lv));
|
|
}
|
|
|
|
if (!arg_count(cmd, yes_ARG) &&
|
|
yes_no_prompt("Do you really want to uncache %s with missing LVs? [y/n]: ",
|
|
display_lvname(lv)) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
if (lvremove_single(cmd, remove_lv, NULL) != ECMD_PROCESSED)
|
|
return_0;
|
|
|
|
if (remove_lv != lv)
|
|
log_print_unless_silent("Logical volume %s is not cached.", display_lvname(lv));
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_snapshot(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
const char *origin_name)
|
|
{
|
|
struct logical_volume *org;
|
|
const char *snap_name;
|
|
uint32_t chunk_size;
|
|
int zero;
|
|
|
|
if (!(snap_name = dm_pool_strdup(lv->vg->vgmem, (display_lvname(lv) ? : ""))))
|
|
return_0;
|
|
|
|
if (strcmp(lv->name, origin_name) == 0) {
|
|
log_error("Unable to use %s as both snapshot and origin.", snap_name);
|
|
return 0;
|
|
}
|
|
|
|
chunk_size = arg_uint_value(cmd, chunksize_ARG, 8);
|
|
if (chunk_size < 8 || chunk_size > 1024 || !is_power_of_2(chunk_size)) {
|
|
log_error("Chunk size must be a power of 2 in the range 4K to 512K.");
|
|
return 0;
|
|
}
|
|
|
|
if (!cow_has_min_chunks(lv->vg, lv->le_count, chunk_size))
|
|
return_0;
|
|
|
|
log_verbose("Setting chunk size to %s.", display_size(cmd, chunk_size));
|
|
|
|
if (!(org = find_lv(lv->vg, origin_name))) {
|
|
log_error("Couldn't find origin volume %s in Volume group %s.",
|
|
origin_name, lv->vg->name);
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* check_lv_rules() checks cannot be done via command definition
|
|
* rules because this LV is not processed by process_each_lv.
|
|
*/
|
|
|
|
/*
|
|
* check_lv_types() checks cannot be done via command definition
|
|
* LV_foo specification because this LV is not processed by process_each_lv.
|
|
*/
|
|
if (!validate_snapshot_origin(org))
|
|
return_0;
|
|
|
|
if (lv_component_is_active(org)) {
|
|
log_error("Cannot use logical volume %s with active component LVs for snapshot origin.",
|
|
display_lvname(org));
|
|
return 0;
|
|
}
|
|
|
|
log_warn("WARNING: Converting logical volume %s to snapshot exception store.",
|
|
snap_name);
|
|
log_warn("THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)");
|
|
|
|
if (!arg_count(cmd, yes_ARG) &&
|
|
yes_no_prompt("Do you really want to convert %s? [y/n]: ",
|
|
snap_name) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
|
|
if (!deactivate_lv(cmd, lv)) {
|
|
log_error("Couldn't deactivate logical volume %s.", snap_name);
|
|
return 0;
|
|
}
|
|
|
|
if (first_seg(lv)->segtype->flags & SEG_CANNOT_BE_ZEROED)
|
|
zero = 0;
|
|
else
|
|
zero = arg_int_value(cmd, zero_ARG, 1);
|
|
|
|
if (!zero || !(lv->status & LVM_WRITE))
|
|
log_warn("WARNING: %s not zeroed.", snap_name);
|
|
else if (!activate_and_wipe_lv(lv, 0)) {
|
|
log_error("Aborting. Failed to wipe snapshot exception store.");
|
|
return 0;
|
|
}
|
|
|
|
if (!archive(lv->vg))
|
|
return_0;
|
|
|
|
if (!vg_add_snapshot(org, lv, NULL, org->le_count, chunk_size)) {
|
|
log_error("Couldn't create snapshot.");
|
|
return 0;
|
|
}
|
|
|
|
/* store vg on disk(s) */
|
|
if (!lv_update_and_reload(org))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Logical volume %s converted to snapshot.", snap_name);
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_merge_old_snapshot(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume **lv_to_poll)
|
|
{
|
|
int merge_on_activate = 0;
|
|
struct logical_volume *origin;
|
|
struct lv_segment *snap_seg = find_snapshot(lv);
|
|
struct lvinfo info;
|
|
dm_percent_t snap_percent;
|
|
|
|
if (!snap_seg)
|
|
return_0;
|
|
|
|
if (!(origin = origin_from_cow(lv))) {
|
|
log_error(INTERNAL_ERROR "Cannot get origin from %s COW.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
/* Check if merge is possible */
|
|
if (lv_is_merging_origin(origin)) {
|
|
log_error("Cannot merge snapshot %s into the origin %s "
|
|
"with merging snapshot %s.",
|
|
display_lvname(lv), display_lvname(origin),
|
|
display_lvname(snap_seg->lv));
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_external_origin(origin)) {
|
|
log_error("Cannot merge snapshot %s into "
|
|
"the read-only external origin %s.",
|
|
display_lvname(lv), display_lvname(origin));
|
|
return 0;
|
|
}
|
|
|
|
if (!(origin->status & LVM_WRITE)) {
|
|
log_error("Cannot merge snapshot %s into "
|
|
"the read-only origin %s. (Use lvchange -p rw).",
|
|
display_lvname(lv), display_lvname(origin));
|
|
return 0;
|
|
}
|
|
|
|
/* FIXME: test when snapshot is remotely active */
|
|
if (lv_info(cmd, lv, 0, &info, 1, 0)
|
|
&& info.exists && info.live_table &&
|
|
(!lv_snapshot_percent(lv, &snap_percent) ||
|
|
snap_percent == DM_PERCENT_INVALID)) {
|
|
log_error("Unable to merge invalidated snapshot LV %s.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (snap_seg->segtype->ops->target_present &&
|
|
!snap_seg->segtype->ops->target_present(cmd, snap_seg, NULL)) {
|
|
log_error("Can't initialize snapshot merge. "
|
|
"Missing support in kernel?");
|
|
return 0;
|
|
}
|
|
|
|
if (!archive(lv->vg))
|
|
return_0;
|
|
|
|
/*
|
|
* Prevent merge with open device(s) as it would likely lead
|
|
* to application/filesystem failure. Merge on origin's next
|
|
* activation if either the origin or snapshot LV are currently
|
|
* open.
|
|
*
|
|
* FIXME testing open_count is racey; snapshot-merge target's
|
|
* constructor and DM should prevent appropriate devices from
|
|
* being open.
|
|
*/
|
|
if (lv_is_active(origin)) {
|
|
if (!lv_check_not_in_use(origin, 0)) {
|
|
log_print_unless_silent("Delaying merge since origin is open.");
|
|
merge_on_activate = 1;
|
|
} else if (!lv_check_not_in_use(lv, 0)) {
|
|
log_print_unless_silent("Delaying merge since snapshot is open.");
|
|
merge_on_activate = 1;
|
|
}
|
|
}
|
|
|
|
init_snapshot_merge(snap_seg, origin);
|
|
|
|
if (merge_on_activate) {
|
|
/* Store and commit vg but skip starting the merge */
|
|
if (!vg_write(lv->vg) || !vg_commit(lv->vg))
|
|
return_0;
|
|
} else {
|
|
/* Perform merge */
|
|
if (!lv_update_and_reload(origin))
|
|
return_0;
|
|
|
|
if (!lv_has_target_type(origin->vg->vgmem, origin, NULL,
|
|
TARGET_NAME_SNAPSHOT_MERGE)) {
|
|
/* Race during table reload prevented merging */
|
|
merge_on_activate = 1;
|
|
|
|
} else if (!lv_is_active(origin)) {
|
|
log_print_unless_silent("Conversion starts after activation.");
|
|
merge_on_activate = 1;
|
|
} else {
|
|
*lv_to_poll = origin;
|
|
}
|
|
}
|
|
|
|
if (merge_on_activate)
|
|
log_print_unless_silent("Merging of snapshot %s will occur on "
|
|
"next activation of %s.",
|
|
display_lvname(lv), display_lvname(origin));
|
|
else
|
|
log_print_unless_silent("Merging of volume %s started.",
|
|
display_lvname(lv));
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_merge_thin_snapshot(struct cmd_context *cmd,
|
|
struct logical_volume *lv)
|
|
{
|
|
int origin_is_active = 0;
|
|
struct lv_segment *snap_seg = first_seg(lv);
|
|
struct logical_volume *origin = snap_seg->origin;
|
|
|
|
if (!origin) {
|
|
log_error("%s is not a mergeable logical volume.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
/* Check if merge is possible */
|
|
if (lv_is_merging_origin(origin)) {
|
|
log_error("Cannot merge snapshot %s into the origin %s "
|
|
"with merging snapshot %s.",
|
|
display_lvname(lv), display_lvname(origin),
|
|
display_lvname(find_snapshot(origin)->lv));
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_external_origin(origin)) {
|
|
if (!(origin = origin_from_cow(lv)))
|
|
log_error(INTERNAL_ERROR "%s is missing origin.",
|
|
display_lvname(lv));
|
|
else
|
|
log_error("%s is read-only external origin %s.",
|
|
display_lvname(lv), display_lvname(origin));
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_origin(origin)) {
|
|
log_error("Merging into the old snapshot origin %s is not supported.",
|
|
display_lvname(origin));
|
|
return 0;
|
|
}
|
|
|
|
if (!archive(lv->vg))
|
|
return_0;
|
|
|
|
/*
|
|
* Prevent merge with open device(s) as it would likely lead
|
|
* to application/filesystem failure. Merge on origin's next
|
|
* activation if either the origin or snapshot LV can't be
|
|
* deactivated.
|
|
*/
|
|
if (!deactivate_lv(cmd, lv))
|
|
log_print_unless_silent("Delaying merge since snapshot is open.");
|
|
else if ((origin_is_active = lv_is_active(origin)) &&
|
|
!deactivate_lv(cmd, origin))
|
|
log_print_unless_silent("Delaying merge since origin volume is open.");
|
|
else {
|
|
/*
|
|
* Both thin snapshot and origin are inactive,
|
|
* replace the origin LV with its snapshot LV.
|
|
*/
|
|
if (!thin_merge_finish(cmd, origin, lv))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Volume %s replaced origin %s.",
|
|
display_lvname(origin), display_lvname(lv));
|
|
|
|
if (origin_is_active && !activate_lv(cmd, lv)) {
|
|
log_error("Failed to reactivate origin %s.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
return 1;
|
|
}
|
|
|
|
init_snapshot_merge(snap_seg, origin);
|
|
|
|
/* Commit vg, merge will start with next activation */
|
|
if (!vg_write(lv->vg) || !vg_commit(lv->vg))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Merging of thin snapshot %s will occur on "
|
|
"next activation of %s.",
|
|
display_lvname(lv), display_lvname(origin));
|
|
return 1;
|
|
}
|
|
|
|
static void _swap_lv_uuid(struct logical_volume *lv1, struct logical_volume *lv2)
|
|
{
|
|
union lvid lvid;
|
|
|
|
if (lv1 && lv2) {
|
|
lvid = lv1->lvid;
|
|
lv1->lvid = lv2->lvid;
|
|
lv2->lvid = lvid;
|
|
}
|
|
}
|
|
|
|
static int _lvconvert_thin_pool_repair(struct cmd_context *cmd,
|
|
struct logical_volume *pool_lv,
|
|
struct dm_list *pvh, int poolmetadataspare)
|
|
{
|
|
const char *thin_dump =
|
|
find_config_tree_str_allow_empty(cmd, global_thin_dump_executable_CFG, NULL);
|
|
int ret = 0, status;
|
|
int args = 0;
|
|
const char *argv[DEFAULT_MAX_EXEC_ARGS + 7] = { /* Max supported args */
|
|
find_config_tree_str_allow_empty(cmd, global_thin_repair_executable_CFG, NULL)
|
|
};
|
|
char *trans_id_str;
|
|
char meta_path[PATH_MAX];
|
|
char pms_path[PATH_MAX];
|
|
uint64_t trans_id;
|
|
struct logical_volume *pmslv;
|
|
struct logical_volume *mlv = first_seg(pool_lv)->metadata_lv;
|
|
struct pipe_data pdata;
|
|
FILE *f;
|
|
|
|
if (!argv[0] || !*argv[0]) {
|
|
log_error("Thin repair command is not configured. Repair is disabled.");
|
|
return 0;
|
|
}
|
|
|
|
if (thin_pool_is_active(pool_lv)) {
|
|
log_error("Cannot repair active pool %s. Use lvchange -an first.",
|
|
display_lvname(pool_lv));
|
|
return 0;
|
|
}
|
|
|
|
pmslv = pool_lv->vg->pool_metadata_spare_lv;
|
|
|
|
/* Check we have pool metadata spare LV */
|
|
if (!handle_pool_metadata_spare(pool_lv->vg, 0, pvh, 1))
|
|
return_0;
|
|
|
|
if (pmslv != pool_lv->vg->pool_metadata_spare_lv) {
|
|
if (!vg_write(pool_lv->vg) || !vg_commit(pool_lv->vg))
|
|
return_0;
|
|
pmslv = pool_lv->vg->pool_metadata_spare_lv;
|
|
}
|
|
|
|
if (dm_snprintf(meta_path, sizeof(meta_path), "%s%s/%s",
|
|
cmd->dev_dir, mlv->vg->name, mlv->name) < 0) {
|
|
log_error("Failed to build thin metadata path.");
|
|
return 0;
|
|
}
|
|
|
|
if (dm_snprintf(pms_path, sizeof(pms_path), "%s%s/%s",
|
|
cmd->dev_dir, pmslv->vg->name, pmslv->name) < 0) {
|
|
log_error("Failed to build pool metadata spare path.");
|
|
return 0;
|
|
}
|
|
|
|
if (!prepare_exec_args(cmd, argv, &args, global_thin_repair_options_CFG))
|
|
return_0;
|
|
|
|
argv[++args] = "-i";
|
|
argv[++args] = meta_path;
|
|
argv[++args] = "-o";
|
|
argv[++args] = pms_path;
|
|
|
|
if (!activate_lv(cmd, pmslv)) {
|
|
log_error("Cannot activate pool metadata spare volume %s.",
|
|
pmslv->name);
|
|
return 0;
|
|
}
|
|
|
|
if (!activate_lv(cmd, mlv)) {
|
|
log_error("Cannot activate thin pool metadata volume %s.",
|
|
mlv->name);
|
|
goto deactivate_pmslv;
|
|
}
|
|
|
|
if (!(ret = exec_cmd(cmd, (const char * const *)argv, &status, 1))) {
|
|
log_error("Repair of thin metadata volume of thin pool %s failed (status:%d). "
|
|
"Manual repair required!",
|
|
display_lvname(pool_lv), status);
|
|
goto deactivate_mlv;
|
|
}
|
|
|
|
/* Check matching transactionId when thin-pool is used by lvm2 (transactionId != 0) */
|
|
if (first_seg(pool_lv)->transaction_id && thin_dump && thin_dump[0]) {
|
|
argv[0] = thin_dump;
|
|
argv[1] = pms_path;
|
|
argv[2] = NULL;
|
|
|
|
if (!(f = pipe_open(cmd, argv, 0, &pdata)))
|
|
log_warn("WARNING: Cannot read output from %s %s.", thin_dump, pms_path);
|
|
else {
|
|
/*
|
|
* Scan only the 1st. line for transation id.
|
|
* Watch out, if the thin_dump format changes
|
|
*/
|
|
if (fgets(meta_path, sizeof(meta_path), f) &&
|
|
(trans_id_str = strstr(meta_path, "transaction=\"")) &&
|
|
(sscanf(trans_id_str + 13, FMTu64, &trans_id) == 1) &&
|
|
(trans_id != first_seg(pool_lv)->transaction_id) &&
|
|
((trans_id - 1) != first_seg(pool_lv)->transaction_id)) {
|
|
log_error("Transaction id " FMTu64 " from pool \"%s/%s\" "
|
|
"does not match repaired transaction id "
|
|
FMTu64 " from %s.",
|
|
first_seg(pool_lv)->transaction_id,
|
|
pool_lv->vg->name, pool_lv->name, trans_id,
|
|
pms_path);
|
|
ret = 0;
|
|
}
|
|
|
|
(void) pipe_close(&pdata); /* killing pipe */
|
|
}
|
|
}
|
|
|
|
deactivate_mlv:
|
|
if (!deactivate_lv(cmd, mlv)) {
|
|
log_error("Cannot deactivate thin pool metadata volume %s.",
|
|
display_lvname(mlv));
|
|
ret = 0;
|
|
}
|
|
|
|
deactivate_pmslv:
|
|
if (!deactivate_lv(cmd, pmslv)) {
|
|
log_error("Cannot deactivate pool metadata spare volume %s.",
|
|
display_lvname(pmslv));
|
|
ret = 0;
|
|
}
|
|
|
|
if (!ret)
|
|
return 0;
|
|
|
|
if (dm_snprintf(meta_path, sizeof(meta_path), "%s_meta%%d", pool_lv->name) < 0) {
|
|
log_error("Can't prepare new metadata name for %s.", pool_lv->name);
|
|
return 0;
|
|
}
|
|
|
|
if (!generate_lv_name(pool_lv->vg, meta_path, pms_path, sizeof(pms_path))) {
|
|
log_error("Can't generate new name for %s.", meta_path);
|
|
return 0;
|
|
}
|
|
|
|
if (pmslv == pool_lv->vg->pool_metadata_spare_lv) {
|
|
pool_lv->vg->pool_metadata_spare_lv = NULL;
|
|
pmslv->status &= ~POOL_METADATA_SPARE;
|
|
lv_set_visible(pmslv);
|
|
}
|
|
|
|
/* Try to allocate new pool metadata spare LV */
|
|
if (!handle_pool_metadata_spare(pool_lv->vg, 0, pvh, poolmetadataspare))
|
|
stack;
|
|
|
|
if (!detach_pool_metadata_lv(first_seg(pool_lv), &mlv))
|
|
return_0;
|
|
|
|
/* TODO: change default to skip */
|
|
lv_set_activation_skip(mlv, 1, arg_int_value(cmd, setactivationskip_ARG, 0));
|
|
mlv->status &= ~LVM_WRITE; /* read-only metadata backup */
|
|
|
|
/* Swap _pmspare and _tmeta name */
|
|
if (!swap_lv_identifiers(cmd, mlv, pmslv))
|
|
return_0;
|
|
|
|
if (!attach_pool_metadata_lv(first_seg(pool_lv), pmslv))
|
|
return_0;
|
|
|
|
/* Used _tmeta (now _pmspare) becomes _meta%d */
|
|
if (!lv_rename_update(cmd, mlv, pms_path, 0))
|
|
return_0;
|
|
|
|
/* Preserve UUID for _pmspare if possible */
|
|
_swap_lv_uuid(mlv, mlv->vg->pool_metadata_spare_lv);
|
|
|
|
if (!vg_write(pool_lv->vg) || !vg_commit(pool_lv->vg))
|
|
return_0;
|
|
|
|
log_warn("WARNING: LV %s holds a backup of the unrepaired metadata. Use lvremove when no longer required.",
|
|
display_lvname(mlv));
|
|
|
|
if (dm_list_size(&pool_lv->vg->pvs) > 1)
|
|
log_warn("WARNING: New metadata LV %s might use different PVs. Move it with pvmove if required.",
|
|
display_lvname(first_seg(pool_lv)->metadata_lv));
|
|
|
|
return 1;
|
|
}
|
|
|
|
/* TODO: lots of similar code with thinpool repair
|
|
* investigate possible better code sharing...
|
|
*/
|
|
static int _lvconvert_cache_repair(struct cmd_context *cmd,
|
|
struct logical_volume *cache_lv,
|
|
struct dm_list *pvh, int poolmetadataspare)
|
|
{
|
|
int ret = 0, status;
|
|
int args = 0;
|
|
const char *argv[DEFAULT_MAX_EXEC_ARGS + 7] = { /* Max supported args */
|
|
find_config_tree_str_allow_empty(cmd, global_cache_repair_executable_CFG, NULL)
|
|
};
|
|
char meta_path[PATH_MAX];
|
|
char pms_path[PATH_MAX];
|
|
struct logical_volume *pool_lv;
|
|
struct logical_volume *pmslv;
|
|
struct logical_volume *mlv;
|
|
|
|
if (lv_is_cache(cache_lv) && lv_is_cache_vol(first_seg(cache_lv)->pool_lv)) {
|
|
log_error("Manual repair required.");
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_active(cache_lv)) {
|
|
log_error("Only inactive cache can be repaired.");
|
|
return 0;
|
|
}
|
|
|
|
pool_lv = lv_is_cache_pool(cache_lv) ? cache_lv : first_seg(cache_lv)->pool_lv;
|
|
mlv = first_seg(pool_lv)->metadata_lv;
|
|
|
|
if (!argv[0] || !*argv[0]) {
|
|
log_error("Cache repair command is not configured. Repair is disabled.");
|
|
return 0; /* Checking disabled */
|
|
}
|
|
|
|
pmslv = cache_lv->vg->pool_metadata_spare_lv;
|
|
|
|
/* Check we have pool metadata spare LV */
|
|
if (!handle_pool_metadata_spare(cache_lv->vg, 0, pvh, 1))
|
|
return_0;
|
|
|
|
if (pmslv != cache_lv->vg->pool_metadata_spare_lv) {
|
|
if (!vg_write(cache_lv->vg) || !vg_commit(cache_lv->vg))
|
|
return_0;
|
|
pmslv = cache_lv->vg->pool_metadata_spare_lv;
|
|
}
|
|
|
|
if (dm_snprintf(meta_path, sizeof(meta_path), "%s%s/%s",
|
|
cmd->dev_dir, mlv->vg->name, mlv->name) < 0) {
|
|
log_error("Failed to build cache metadata path.");
|
|
return 0;
|
|
}
|
|
|
|
if (dm_snprintf(pms_path, sizeof(pms_path), "%s%s/%s",
|
|
cmd->dev_dir, pmslv->vg->name, pmslv->name) < 0) {
|
|
log_error("Failed to build pool metadata spare path.");
|
|
return 0;
|
|
}
|
|
|
|
if (!prepare_exec_args(cmd, argv, &args, global_cache_repair_options_CFG))
|
|
return_0;
|
|
|
|
argv[++args] = "-i";
|
|
argv[++args] = meta_path;
|
|
argv[++args] = "-o";
|
|
argv[++args] = pms_path;
|
|
|
|
if (!activate_lv(cmd, pmslv)) {
|
|
log_error("Cannot activate pool metadata spare volume %s.",
|
|
pmslv->name);
|
|
return 0;
|
|
}
|
|
|
|
if (!activate_lv(cmd, mlv)) {
|
|
log_error("Cannot activate cache pool metadata volume %s.",
|
|
mlv->name);
|
|
goto deactivate_pmslv;
|
|
}
|
|
|
|
if (!(ret = exec_cmd(cmd, (const char * const *)argv, &status, 1))) {
|
|
log_error("Repair of cache metadata volume of cache %s failed (status:%d). "
|
|
"Manual repair required!",
|
|
display_lvname(cache_lv), status);
|
|
goto deactivate_mlv;
|
|
}
|
|
|
|
/* TODO: any active validation of cache-pool metadata? */
|
|
|
|
deactivate_mlv:
|
|
if (!deactivate_lv(cmd, mlv)) {
|
|
log_error("Cannot deactivate pool metadata volume %s.",
|
|
display_lvname(mlv));
|
|
ret = 0;
|
|
}
|
|
|
|
deactivate_pmslv:
|
|
if (!deactivate_lv(cmd, pmslv)) {
|
|
log_error("Cannot deactivate pool metadata spare volume %s.",
|
|
display_lvname(pmslv));
|
|
ret = 0;
|
|
}
|
|
|
|
if (!ret)
|
|
return 0;
|
|
|
|
if (dm_snprintf(meta_path, sizeof(meta_path), "%s_meta%%d", pool_lv->name) < 0) {
|
|
log_error("Can't prepare new metadata name for %s.", display_lvname(pool_lv));
|
|
return 0;
|
|
}
|
|
|
|
if (!generate_lv_name(cache_lv->vg, meta_path, pms_path, sizeof(pms_path))) {
|
|
log_error("Can't generate new name for %s.", meta_path);
|
|
return 0;
|
|
}
|
|
|
|
if (pmslv == cache_lv->vg->pool_metadata_spare_lv) {
|
|
cache_lv->vg->pool_metadata_spare_lv = NULL;
|
|
pmslv->status &= ~POOL_METADATA_SPARE;
|
|
lv_set_visible(pmslv);
|
|
}
|
|
|
|
/* Try to allocate new pool metadata spare LV */
|
|
if (!handle_pool_metadata_spare(cache_lv->vg, 0, pvh, poolmetadataspare))
|
|
stack;
|
|
|
|
if (!detach_pool_metadata_lv(first_seg(pool_lv), &mlv))
|
|
return_0;
|
|
|
|
/* TODO: change default to skip */
|
|
lv_set_activation_skip(mlv, 1, arg_int_value(cmd, setactivationskip_ARG, 0));
|
|
mlv->status &= ~LVM_WRITE; /* read-only metadata backup */
|
|
|
|
/* Swap _pmspare and _cmeta name */
|
|
if (!swap_lv_identifiers(cmd, mlv, pmslv))
|
|
return_0;
|
|
|
|
if (!attach_pool_metadata_lv(first_seg(pool_lv), pmslv))
|
|
return_0;
|
|
|
|
/* Used _cmeta (now _pmspare) becomes _meta%d */
|
|
if (!lv_rename_update(cmd, mlv, pms_path, 0))
|
|
return_0;
|
|
|
|
/* Preserve UUID for _pmspare if possible */
|
|
_swap_lv_uuid(mlv, mlv->vg->pool_metadata_spare_lv);
|
|
|
|
if (!vg_write(cache_lv->vg) || !vg_commit(cache_lv->vg))
|
|
return_0;
|
|
|
|
/* FIXME: just as with thinpool repair - fix the warning
|
|
* where moving doesn't make any sense (same disk storage)
|
|
*/
|
|
log_warn("WARNING: If everything works, remove %s volume.",
|
|
display_lvname(mlv));
|
|
|
|
log_warn("WARNING: Use pvmove command to move %s on the best fitting PV.",
|
|
display_lvname(first_seg(pool_lv)->metadata_lv));
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_to_thin_with_external(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *thinpool_lv)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *thin_lv;
|
|
const char *origin_name;
|
|
int lv_was_active;
|
|
struct lvcreate_params lvc = {
|
|
.activate = CHANGE_AEY,
|
|
.alloc = ALLOC_INHERIT,
|
|
.major = -1,
|
|
.minor = -1,
|
|
.suppress_zero_warn = 1, /* Suppress warning for this thin */
|
|
.permission = LVM_READ,
|
|
.pool_name = thinpool_lv->name,
|
|
.pvh = &vg->pvs,
|
|
.read_ahead = DM_READ_AHEAD_AUTO,
|
|
.stripes = 1,
|
|
.virtual_extents = lv->le_count,
|
|
.tags = DM_LIST_HEAD_INIT(lvc.tags),
|
|
};
|
|
|
|
if (!_raid_split_image_conversion(lv))
|
|
return_0;
|
|
|
|
if (lv == thinpool_lv) {
|
|
log_error("Can't use same LV %s for thin pool and thin volume.",
|
|
display_lvname(thinpool_lv));
|
|
return 0;
|
|
}
|
|
|
|
if ((origin_name = arg_str_value(cmd, originname_ARG, NULL)))
|
|
if (!validate_restricted_lvname_param(cmd, &vg->name, &origin_name))
|
|
return_0;
|
|
|
|
/*
|
|
* If NULL, an auto-generated 'lvol' name is used.
|
|
* If set, the lv create code checks the name isn't used.
|
|
*/
|
|
lvc.lv_name = origin_name;
|
|
|
|
if (vg_is_shared(vg)) {
|
|
/*
|
|
* FIXME: external origins don't work in lockd VGs.
|
|
* Prior to the lvconvert, there's a lock associated with
|
|
* the uuid of the external origin LV. After the convert,
|
|
* that uuid belongs to the new thin LV, and a new LV with
|
|
* a new uuid exists as the non-thin, readonly external LV.
|
|
* We'd need to remove the lock for the previous uuid
|
|
* (the new thin LV will have no lock), and create a new
|
|
* lock for the new LV uuid used by the external LV.
|
|
*/
|
|
log_error("Can't use lock_type %s LV as external origin.",
|
|
vg->lock_type);
|
|
return 0;
|
|
}
|
|
|
|
if (!thin_pool_supports_external_origin(first_seg(thinpool_lv), lv))
|
|
return_0;
|
|
|
|
if (!(lvc.segtype = get_segtype_from_string(cmd, SEG_TYPE_NAME_THIN)))
|
|
return_0;
|
|
|
|
lv_was_active = lv_is_active(lv);
|
|
|
|
/* When converted LV is not holding lock, but some other LV keeps it
|
|
* 'active' i.e. being an external origin for such LV, activate this LV
|
|
* so the reload of table can properly update device tree. */
|
|
if (!lv_was_active && (lv != lv_lock_holder(lv)) && !activate_lv(cmd, lv)) {
|
|
log_error("Failed to activate %s. Conversion cannot proceed.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* New thin LV needs to be created (all messages sent to pool) In this
|
|
* case thin volume is created READ-ONLY and also warn about not
|
|
* zeroing is suppressed.
|
|
*
|
|
* The new thin LV is created with the origin_name, or an autogenerated
|
|
* 'lvol' name. Then the names and ids are swapped between the thin LV
|
|
* and the original/external LV. So, the thin LV gets the name and id
|
|
* of the original LV arg, and the original LV arg gets the origin_name
|
|
* or the autogenerated name.
|
|
*/
|
|
|
|
if (!(thin_lv = lv_create_single(vg, &lvc)))
|
|
return_0;
|
|
|
|
/*
|
|
* Only for converted active thick snapshot origin leave
|
|
* created thin LV active (locked) so it can be converted to new
|
|
* read-only 'snapshot-origin' with the consequent update and reload.
|
|
*
|
|
* Note: New thin LV is read-only so it can't be written.
|
|
*/
|
|
if (!lv_is_origin(lv) || !lv_was_active) {
|
|
if (!deactivate_lv(cmd, thin_lv)) {
|
|
log_error("Aborting. Failed to deactivate new thin LV. "
|
|
"Manual intervention required.");
|
|
return 0;
|
|
}
|
|
if (!sync_local_dev_names(cmd)) {
|
|
log_error("Failed to sync local devices before conversion.");
|
|
goto revert_new_lv;
|
|
}
|
|
}
|
|
|
|
/*
|
|
* Crashing till this point will leave plain thin volume
|
|
* which could be easily removed by the user after i.e. power-off
|
|
*/
|
|
if (!swap_lv_identifiers(cmd, thin_lv, lv)) {
|
|
log_error("Aborting. Failed to swap identifiers. "
|
|
"Manual intervention required.");
|
|
return 0; /* runtime corruption */
|
|
}
|
|
|
|
/* Preserve read-write status of original LV here */
|
|
thin_lv->status |= (lv->status & LVM_WRITE);
|
|
|
|
if (!attach_thin_external_origin(first_seg(thin_lv), lv)) {
|
|
log_error("Aborting. Failed to attach external origin. "
|
|
"Manual intervention required.");
|
|
return 0; /* runtime corruption */
|
|
}
|
|
|
|
if (!lv_update_and_reload(thin_lv)) {
|
|
stack;
|
|
goto deactivate_and_revert_new_lv;
|
|
}
|
|
|
|
log_print_unless_silent("Converted %s to thin volume with external origin %s.",
|
|
display_lvname(thin_lv), display_lvname(lv));
|
|
|
|
/* Restore previous state */
|
|
if (!lv_was_active && !deactivate_lv(cmd, thin_lv)) {
|
|
log_error("Failed to deactivate thin LV %s.", display_lvname(thin_lv));
|
|
return 0;
|
|
}
|
|
|
|
return 1;
|
|
|
|
deactivate_and_revert_new_lv:
|
|
if (!detach_thin_external_origin(first_seg(thin_lv)))
|
|
return_0;
|
|
|
|
if (!swap_lv_identifiers(cmd, thin_lv, lv))
|
|
return_0;
|
|
|
|
if (!deactivate_lv(cmd, thin_lv)) {
|
|
log_error("Failed to deactivate thin LV. "
|
|
"Manual intervention required.");
|
|
return 0;
|
|
}
|
|
|
|
revert_new_lv:
|
|
/* FIXME Better to revert to backup of metadata? */
|
|
if (!lv_remove(thin_lv) || !vg_write(vg) || !vg_commit(vg))
|
|
log_error("Manual intervention may be required to remove "
|
|
"abandoned LV(s) before retrying.");
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int _lvconvert_swap_pool_metadata(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *metadata_lv)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *prev_metadata_lv;
|
|
struct lv_segment *seg;
|
|
const struct lv_type *lvtype;
|
|
char meta_name[NAME_LEN];
|
|
const char *swap_lock_args = NULL;
|
|
uint32_t chunk_size;
|
|
int is_thinpool;
|
|
int is_cachepool;
|
|
int lvt_enum;
|
|
|
|
is_thinpool = lv_is_thin_pool(lv);
|
|
is_cachepool = lv_is_cache_pool(lv);
|
|
lvt_enum = get_lvt_enum(metadata_lv);
|
|
lvtype = get_lv_type(lvt_enum);
|
|
|
|
if (lvt_enum != striped_LVT && lvt_enum != linear_LVT && lvt_enum != raid_LVT) {
|
|
log_error("LV %s with type %s cannot be used as a metadata LV.",
|
|
display_lvname(metadata_lv), lvtype ? lvtype->name : "unknown");
|
|
return 0;
|
|
}
|
|
|
|
if (!lv_is_visible(metadata_lv)) {
|
|
log_error("Can't convert internal LV %s.",
|
|
display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_locked(metadata_lv)) {
|
|
log_error("Can't convert locked LV %s.",
|
|
display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_origin(metadata_lv) ||
|
|
lv_is_merging_origin(metadata_lv) ||
|
|
lv_is_external_origin(metadata_lv) ||
|
|
lv_is_virtual(metadata_lv)) {
|
|
log_error("Pool metadata LV %s is of an unsupported type.",
|
|
display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
/* FIXME cache pool */
|
|
if (is_thinpool && thin_pool_is_active(lv)) {
|
|
/* If any volume referencing pool active - abort here */
|
|
log_error("Cannot convert pool %s with active volumes.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if ((dm_snprintf(meta_name, sizeof(meta_name), "%s%s", lv->name, is_cachepool ? "_cmeta" : "_tmeta") < 0)) {
|
|
log_error("Failed to create internal lv names, pool name is too long.");
|
|
return 0;
|
|
}
|
|
|
|
/* If LV is inactive here, ensure it's not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
return 0;
|
|
|
|
/* If new metadata LV is inactive here, ensure it's not active elsewhere */
|
|
if (!lockd_lv(cmd, metadata_lv, "ex", 0)) {
|
|
log_error("New pool metadata LV %s cannot be locked.", display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
if (!deactivate_lv(cmd, metadata_lv)) {
|
|
log_error("Aborting. Failed to deactivate %s.",
|
|
display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* metadata_lv is currently an independent LV with its own lockd lock allocated.
|
|
* A pool metadata LV does not have its own lockd lock (only the pool LV does.)
|
|
* Since the LV name and uuid are exchanged between the old and new metadata LVs,
|
|
* the lvmlockd lock can just be moved between the two LVs, so the new indepdent
|
|
* LV (former metadata LV) gets the lock that was used for old independent LV.
|
|
*/
|
|
if (vg_is_shared(vg) && metadata_lv->lock_args) {
|
|
swap_lock_args = metadata_lv->lock_args;
|
|
metadata_lv->lock_args = NULL;
|
|
}
|
|
|
|
seg = first_seg(lv);
|
|
|
|
/* Normally do NOT change chunk size when swapping */
|
|
|
|
if (arg_is_set(cmd, chunksize_ARG)) {
|
|
chunk_size = arg_uint_value(cmd, chunksize_ARG, 0);
|
|
|
|
if ((chunk_size != seg->chunk_size) && !dm_list_empty(&lv->segs_using_this_lv)) {
|
|
if (arg_count(cmd, force_ARG) == PROMPT) {
|
|
log_error("Chunk size can be only changed with --force. Conversion aborted.");
|
|
return 0;
|
|
}
|
|
|
|
if (!validate_pool_chunk_size(cmd, seg->segtype, chunk_size))
|
|
return_0;
|
|
|
|
log_warn("WARNING: Changing chunk size %s to %s for %s pool volume.",
|
|
display_size(cmd, seg->chunk_size),
|
|
display_size(cmd, chunk_size),
|
|
display_lvname(lv));
|
|
|
|
/* Ok, user has likely some serious reason for this */
|
|
if (!arg_count(cmd, yes_ARG) &&
|
|
yes_no_prompt("Do you really want to change chunk size for %s pool volume? [y/n]: ",
|
|
display_lvname(lv)) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
seg->chunk_size = chunk_size;
|
|
}
|
|
|
|
if (!arg_count(cmd, yes_ARG) &&
|
|
yes_no_prompt("Do you want to swap metadata of %s pool with metadata volume %s? [y/n]: ",
|
|
display_lvname(lv),
|
|
display_lvname(metadata_lv)) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
|
|
/* Swap names between old and new metadata LV */
|
|
|
|
if (!detach_pool_metadata_lv(seg, &prev_metadata_lv))
|
|
return_0;
|
|
|
|
if (!swap_lv_identifiers(cmd, metadata_lv, prev_metadata_lv))
|
|
return_0;
|
|
|
|
if (!attach_pool_metadata_lv(seg, metadata_lv))
|
|
return_0;
|
|
|
|
/*
|
|
* The previous metadata LV will now be an independent LV so it now
|
|
* requires a lockd lock, and gets the lock from the LV that's becoming
|
|
* the new metadata LV.
|
|
*/
|
|
prev_metadata_lv->lock_args = swap_lock_args;
|
|
|
|
if (!vg_write(vg) || !vg_commit(vg))
|
|
return_0;
|
|
|
|
return 1;
|
|
}
|
|
|
|
static struct logical_volume *_lvconvert_insert_thin_layer(struct logical_volume *lv)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct segment_type *thin_segtype;
|
|
struct logical_volume *pool_lv;
|
|
struct lv_segment *seg;
|
|
|
|
if (!(thin_segtype = get_segtype_from_string(vg->cmd, SEG_TYPE_NAME_THIN)))
|
|
return_NULL;
|
|
|
|
/*
|
|
* input lv foo (often linear)
|
|
* creates new lv foo_tpoolN (no seg)
|
|
* segment from foo is moved to foo_tpoolN
|
|
* new linear segment is created for foo that maps to foo_tpoolN
|
|
* returns foo_tpoolN
|
|
*
|
|
* In spite of the "pool" variable naming, pool_lv foo_tpoolN is *not*
|
|
* yet a pool type, but rather is whatever type the input lv was.
|
|
*/
|
|
if (!(pool_lv = insert_layer_for_lv(vg->cmd, lv, 0, "_tpool%d")))
|
|
return_NULL;
|
|
|
|
/*
|
|
* change lv foo to a thin LV using foo_tpoolN
|
|
*/
|
|
lv->status |= THIN_VOLUME | VIRTUAL;
|
|
lv_set_visible(pool_lv);
|
|
|
|
seg = first_seg(lv);
|
|
seg->area_count = 0;
|
|
|
|
seg->segtype = thin_segtype;
|
|
seg->pool_lv = pool_lv;
|
|
seg->device_id = 1;
|
|
seg->transaction_id = 0;
|
|
|
|
return pool_lv;
|
|
}
|
|
|
|
/*
|
|
* Create a new pool LV, using the lv arg as the data sub LV.
|
|
* The metadata sub LV is either a new LV created here, or an
|
|
* existing LV specified by --poolmetadata.
|
|
*
|
|
* process_single_lv is the LV currently being processed by
|
|
* process_each_lv(). It will sometimes be the same as the
|
|
* lv arg, and sometimes not.
|
|
*/
|
|
|
|
static int _lvconvert_to_pool(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *process_single_lv,
|
|
int to_thinpool,
|
|
int to_cachepool,
|
|
int to_thin,
|
|
struct dm_list *use_pvh)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *metadata_lv = NULL; /* existing or created */
|
|
struct logical_volume *data_lv; /* lv arg renamed */
|
|
struct logical_volume *pool_lv = NULL; /* new lv created here */
|
|
const char *pool_metadata_name; /* user-specified lv name */
|
|
char converted_names[3*NAME_LEN] = { 0 }; /* preserve names of converted lv */
|
|
struct segment_type *pool_segtype; /* thinpool or cachepool */
|
|
const char *str_seg_type = to_cachepool ? SEG_TYPE_NAME_CACHE_POOL : SEG_TYPE_NAME_THIN_POOL;
|
|
struct lv_segment *seg;
|
|
unsigned int target_attr = ~0;
|
|
unsigned int activate_pool;
|
|
unsigned int lock_active_pool_done = 0;
|
|
unsigned int zero_metadata;
|
|
uint64_t meta_size;
|
|
uint32_t meta_extents;
|
|
uint32_t chunk_size;
|
|
int chunk_calc;
|
|
cache_metadata_format_t cache_metadata_format;
|
|
cache_mode_t cache_mode;
|
|
const char *policy_name;
|
|
struct dm_config_tree *policy_settings = NULL;
|
|
int pool_metadata_spare;
|
|
thin_crop_metadata_t crop_metadata;
|
|
thin_discards_t discards;
|
|
thin_zero_t zero_new_blocks;
|
|
int error_when_full;
|
|
int data_vdo;
|
|
uint64_t vdo_pool_header_size;
|
|
struct vdo_convert_params vcp = {
|
|
.activate = CHANGE_AN,
|
|
.do_zero = 1,
|
|
.do_wipe_signatures = 1,
|
|
.force = arg_count(cmd, force_ARG),
|
|
.yes = arg_count(cmd, yes_ARG),
|
|
};
|
|
int is_active;
|
|
int ret = 1;
|
|
|
|
/* for handling lvmlockd cases */
|
|
char *lockd_data_args = NULL;
|
|
char *lockd_meta_args = NULL;
|
|
char *lockd_data_name = NULL;
|
|
char *lockd_meta_name = NULL;
|
|
uint32_t lockd_data_flags = 0;
|
|
uint32_t lockd_meta_flags = 0;
|
|
struct id lockd_data_id;
|
|
struct id lockd_meta_id;
|
|
|
|
if (!_raid_split_image_conversion(lv))
|
|
return_0;
|
|
|
|
if (lv_is_thin_pool(lv) || lv_is_cache_pool(lv)) {
|
|
log_error(INTERNAL_ERROR "LV %s is already a pool.", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
pool_segtype = get_segtype_from_string(cmd, str_seg_type);
|
|
|
|
if (!pool_segtype ||
|
|
!pool_segtype->ops->target_present(cmd, NULL, &target_attr)) {
|
|
log_error("%s: Required device-mapper target(s) not detected in your kernel.",
|
|
str_seg_type);
|
|
return 0;
|
|
}
|
|
|
|
/* If LV is inactive here, ensure it's not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
return 0;
|
|
|
|
is_active = lv_is_active(lv);
|
|
|
|
activate_pool = to_thinpool && is_active;
|
|
|
|
/* Wipe metadata_lv by default, but allow skipping this for cache pools. */
|
|
zero_metadata = (to_cachepool) ? arg_int_value(cmd, zero_ARG, 1) : 1;
|
|
|
|
/*
|
|
* If an existing LV is to be used as the metadata LV,
|
|
* verify that it's in a usable state. These checks are
|
|
* not done by command def rules because this LV is not
|
|
* processed by process_each_lv.
|
|
*/
|
|
|
|
if ((pool_metadata_name = arg_str_value(cmd, poolmetadata_ARG, NULL))) {
|
|
if (!validate_lvname_param(cmd, &vg->name, &pool_metadata_name)) {
|
|
log_error("Metadata LV %s not found.", pool_metadata_name);
|
|
return 0;
|
|
}
|
|
|
|
if (!(metadata_lv = find_lv(vg, pool_metadata_name))) {
|
|
log_error("Unknown pool metadata LV %s.", pool_metadata_name);
|
|
return 0;
|
|
}
|
|
|
|
if (metadata_lv == lv) {
|
|
log_error("Can't use same LV for pool data and metadata LV %s.",
|
|
display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
if (metadata_lv == process_single_lv) {
|
|
log_error("Use a different LV for pool metadata %s.",
|
|
display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
if (!lv_is_visible(metadata_lv)) {
|
|
log_error("Can't convert internal LV %s.",
|
|
display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_locked(metadata_lv)) {
|
|
log_error("Can't convert locked LV %s.",
|
|
display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
if (lv_is_mirror(metadata_lv)) {
|
|
log_error("Mirror logical volumes cannot be used for pool metadata.");
|
|
log_print_unless_silent("Try \"%s\" segment type instead.", SEG_TYPE_NAME_RAID1);
|
|
return 0;
|
|
}
|
|
|
|
/* FIXME Tidy up all these type restrictions. (Use a type whitelist?) */
|
|
if (lv_is_cache_type(metadata_lv) ||
|
|
lv_is_writecache(metadata_lv) ||
|
|
lv_is_thin_type(metadata_lv) ||
|
|
lv_is_cow(metadata_lv) || lv_is_merging_cow(metadata_lv) ||
|
|
lv_is_origin(metadata_lv) || lv_is_merging_origin(metadata_lv) ||
|
|
lv_is_external_origin(metadata_lv) ||
|
|
lv_is_virtual(metadata_lv)) {
|
|
log_error("Pool metadata LV %s is of an unsupported type.",
|
|
display_lvname(metadata_lv));
|
|
return 0;
|
|
}
|
|
|
|
/* If LV is inactive here, ensure it's not active elsewhere. */
|
|
if (!lockd_lv(cmd, metadata_lv, "ex", 0))
|
|
return 0;
|
|
|
|
/* An existing LV needs to have its lock freed once it becomes a meta LV. */
|
|
if (vg_is_shared(vg) && metadata_lv->lock_args) {
|
|
lockd_meta_args = dm_pool_strdup(vg->vgmem, metadata_lv->lock_args);
|
|
lockd_meta_name = dm_pool_strdup(vg->vgmem, metadata_lv->name);
|
|
lockd_meta_flags = lv_is_active(metadata_lv) ? LDLV_PERSISTENT : 0;
|
|
lockd_meta_id = metadata_lv->lvid.id[1];
|
|
}
|
|
}
|
|
|
|
if (vg_is_shared(vg) && lv->lock_args) {
|
|
lockd_data_args = dm_pool_strdup(vg->vgmem, lv->lock_args);
|
|
lockd_data_name = dm_pool_strdup(vg->vgmem, lv->name);
|
|
lockd_data_flags = is_active ? LDLV_PERSISTENT : 0;
|
|
lockd_data_id = lv->lvid.id[1];
|
|
}
|
|
|
|
if (!get_pool_params(cmd, pool_segtype,
|
|
&data_vdo, &meta_size, &pool_metadata_spare,
|
|
&chunk_size, &discards, &zero_new_blocks))
|
|
goto_bad;
|
|
|
|
if (to_cachepool &&
|
|
!get_cache_params(cmd, &chunk_size, &cache_metadata_format, &cache_mode, &policy_name, &policy_settings))
|
|
goto_bad;
|
|
|
|
if (metadata_lv)
|
|
meta_extents = metadata_lv->le_count;
|
|
else if (meta_size)
|
|
meta_extents = extents_from_size(cmd, meta_size, vg->extent_size);
|
|
else
|
|
meta_extents = 0; /* A default will be chosen by the "update" function. */
|
|
|
|
/*
|
|
* Validate and/or choose defaults for meta_extents and chunk_size,
|
|
* this involves some complicated calculations.
|
|
*/
|
|
|
|
if (to_cachepool) {
|
|
if (!update_cache_pool_params(cmd, vg->profile, vg->extent_size,
|
|
pool_segtype, target_attr,
|
|
lv->le_count,
|
|
&meta_extents,
|
|
metadata_lv,
|
|
&chunk_calc,
|
|
&chunk_size))
|
|
goto_bad;
|
|
} else {
|
|
if (!update_thin_pool_params(cmd, vg->profile, vg->extent_size,
|
|
pool_segtype, target_attr,
|
|
lv->le_count,
|
|
&meta_extents,
|
|
metadata_lv,
|
|
&crop_metadata,
|
|
&chunk_calc,
|
|
&chunk_size,
|
|
&discards, &zero_new_blocks))
|
|
goto_bad;
|
|
}
|
|
|
|
if (metadata_lv && (meta_extents > metadata_lv->le_count)) {
|
|
log_error("Pool metadata LV %s is too small (%u extents) for required metadata (%u extents).",
|
|
display_lvname(metadata_lv), metadata_lv->le_count, meta_extents);
|
|
goto bad;
|
|
}
|
|
|
|
log_verbose("Pool metadata extents %u chunk_size %u", meta_extents, chunk_size);
|
|
|
|
(void) dm_snprintf(converted_names, sizeof(converted_names), "%s%s%s",
|
|
display_lvname(lv),
|
|
metadata_lv ? " and " : "",
|
|
metadata_lv ? display_lvname(metadata_lv) : "");
|
|
|
|
/* Verify user really wants to convert these LVs. */
|
|
if (!to_thin)
|
|
log_warn("WARNING: Converting %s to %s pool's data%s %s metadata wiping.",
|
|
converted_names,
|
|
to_cachepool ? "cache" : "thin",
|
|
metadata_lv ? " and metadata volumes" : " volume",
|
|
zero_metadata ? "with" : "WITHOUT");
|
|
|
|
if (to_thin)
|
|
log_warn("WARNING: Converting %s to fully provisioned thin volume.",
|
|
converted_names);
|
|
else if (zero_metadata) {
|
|
if (lv_is_error(lv) || lv_is_zero(lv))
|
|
log_warn("WARNING: Volume of \"%s\" segtype cannot store ANY real data!",
|
|
first_seg(lv)->segtype->name);
|
|
else
|
|
log_warn("THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)");
|
|
} else if (to_cachepool)
|
|
log_warn("WARNING: Using mismatched cache pool metadata MAY DESTROY YOUR DATA!");
|
|
|
|
if (!arg_count(cmd, yes_ARG) &&
|
|
yes_no_prompt("Do you really want to convert %s? [y/n]: ",
|
|
converted_names) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
goto bad;
|
|
}
|
|
|
|
/*
|
|
* If a new metadata LV needs to be created, collect the settings for
|
|
* the new LV and create it.
|
|
*
|
|
* If an existing LV is used for metadata, deactivate/activate/wipe it.
|
|
*/
|
|
|
|
if (!metadata_lv) {
|
|
uint32_t meta_stripes;
|
|
uint32_t meta_stripe_size;
|
|
uint32_t meta_readahead;
|
|
alloc_policy_t meta_alloc;
|
|
unsigned meta_stripes_supplied;
|
|
unsigned meta_stripe_size_supplied;
|
|
|
|
if (!get_stripe_params(cmd, get_segtype_from_string(cmd, SEG_TYPE_NAME_STRIPED),
|
|
&meta_stripes,
|
|
&meta_stripe_size,
|
|
&meta_stripes_supplied,
|
|
&meta_stripe_size_supplied))
|
|
goto_bad;
|
|
|
|
meta_readahead = arg_uint_value(cmd, readahead_ARG, cmd->default_settings.read_ahead);
|
|
meta_alloc = (alloc_policy_t) arg_uint_value(cmd, alloc_ARG, ALLOC_INHERIT);
|
|
|
|
if (!(metadata_lv = alloc_pool_metadata(lv,
|
|
meta_readahead,
|
|
meta_stripes,
|
|
meta_stripe_size,
|
|
meta_extents,
|
|
meta_alloc,
|
|
use_pvh)))
|
|
goto_bad;
|
|
} else {
|
|
if (!deactivate_lv(cmd, metadata_lv)) {
|
|
log_error("Aborting. Failed to deactivate %s.",
|
|
display_lvname(metadata_lv));
|
|
goto bad;
|
|
}
|
|
|
|
if (zero_metadata || to_thin) {
|
|
metadata_lv->status |= LV_ACTIVATION_SKIP;
|
|
if (!activate_lv(cmd, metadata_lv)) {
|
|
log_error("Aborting. Failed to activate metadata lv.");
|
|
goto bad;
|
|
}
|
|
metadata_lv->status &= ~LV_ACTIVATION_SKIP;
|
|
|
|
if (!wipe_lv(metadata_lv, (struct wipe_params) {
|
|
.do_wipe_signatures = 1,
|
|
.is_metadata = 1,
|
|
.yes = arg_count(cmd, yes_ARG),
|
|
.force = arg_count(cmd, force_ARG) } )) {
|
|
log_error("Aborting. Failed to wipe metadata lv.");
|
|
goto bad;
|
|
}
|
|
}
|
|
}
|
|
|
|
/*
|
|
* Before starting a real conversion, prepare _pmspare volume.
|
|
* If there is already one presend in a VG, make sure the size is right
|
|
*/
|
|
if (!handle_pool_metadata_spare(vg, metadata_lv->le_count, use_pvh, pool_metadata_spare)) {
|
|
log_error("Failed to set up spare metadata LV for pool.");
|
|
goto bad;
|
|
}
|
|
/*
|
|
* After _pmspare handling is finished there are unwritten VG metadata
|
|
* that will get written with the next mda update.
|
|
* If there is any failure and such new metadata would not be written,
|
|
* a user is then left with 'regular' volume he can normally 'lvremove'.
|
|
* If there was _pmspare already existing, the size of such LV is already changed
|
|
* and is committed to disk. A user may only remove such volume and create a new one.
|
|
*/
|
|
|
|
if (to_thin) {
|
|
/*
|
|
* pool_lv is not yet a pool, when returned, pool_lv contains
|
|
* the segment that belonged to "lv".
|
|
*/
|
|
if (!(pool_lv = _lvconvert_insert_thin_layer(lv)))
|
|
goto_bad;
|
|
} else {
|
|
/* Deactivate the data LV (changing target type) */
|
|
if (!deactivate_lv(cmd, lv)) {
|
|
log_error("Aborting. Failed to deactivate logical volume %s.",
|
|
display_lvname(lv));
|
|
goto bad;
|
|
}
|
|
|
|
if (data_vdo) {
|
|
if (!fill_vdo_target_params(cmd, &vcp.vdo_params, &vdo_pool_header_size, vg->profile))
|
|
goto_bad;
|
|
|
|
if (!get_vdo_settings(cmd, &vcp.vdo_params, NULL))
|
|
goto_bad;
|
|
|
|
if (data_vdo && lv_is_vdo(lv))
|
|
log_print_unless_silent("Volume %s is already VDO volume, skipping VDO conversion.",
|
|
display_lvname(lv));
|
|
else if (!convert_vdo_lv(lv, &vcp))
|
|
goto_bad;
|
|
}
|
|
|
|
pool_lv = lv;
|
|
}
|
|
|
|
/*
|
|
* starts with pool_lv foo (not a pool yet)
|
|
* creates new data_lv foo_tdata
|
|
* segment from pool_lv foo is moved to data_lv foo_tdata
|
|
* pool_lv foo linear segment is created that maps to foo_tdata
|
|
* returns data_lv foo_tdata
|
|
*
|
|
* (In the to_thin case, the segment from the original lv is first
|
|
* moved to pool_lv by _lvconvert_insert_thin_layer, and now is
|
|
* moved to data_lv.)
|
|
*/
|
|
if (!(data_lv = insert_layer_for_lv(cmd, pool_lv, 0,
|
|
(to_cachepool ? "_cdata" : "_tdata"))))
|
|
goto_bad;
|
|
|
|
data_lv->status |= (to_cachepool) ? CACHE_POOL_DATA : THIN_POOL_DATA;
|
|
data_lv->status |= LVM_WRITE; /* Pool data LV is writable */
|
|
|
|
/*
|
|
* pool_lv now becomes a pool type.
|
|
* FIXME: change variable naming to avoid this confusion.
|
|
*/
|
|
pool_lv->status |= (to_cachepool) ? CACHE_POOL : THIN_POOL;
|
|
|
|
seg = first_seg(pool_lv);
|
|
seg->segtype = pool_segtype;
|
|
|
|
/* Apply settings to the new pool seg */
|
|
if (to_cachepool) {
|
|
if (!cache_set_params(seg, chunk_size, cache_metadata_format, cache_mode, policy_name, policy_settings))
|
|
goto_bad;
|
|
} else {
|
|
/* Error when full */
|
|
if (arg_is_set(cmd, errorwhenfull_ARG))
|
|
error_when_full = arg_int_value(cmd, errorwhenfull_ARG, 0);
|
|
else
|
|
error_when_full = find_config_tree_bool(cmd, activation_error_when_full_CFG, vg->profile);
|
|
|
|
if (!thin_pool_set_params(seg,
|
|
error_when_full,
|
|
crop_metadata,
|
|
chunk_calc,
|
|
chunk_size,
|
|
discards,
|
|
zero_new_blocks))
|
|
goto_bad;
|
|
|
|
if (to_thin) {
|
|
if (!thin_pool_prepare_metadata(metadata_lv, seg->chunk_size,
|
|
pool_lv->size / seg->chunk_size,
|
|
0,
|
|
pool_lv->size / seg->chunk_size))
|
|
goto_bad;
|
|
seg->transaction_id = 1;
|
|
}
|
|
}
|
|
|
|
if (!add_metadata_to_pool(seg, metadata_lv))
|
|
goto_bad;
|
|
|
|
/*
|
|
* If the input LV is being converted to a thin pool, the input LV lock
|
|
* is used for the thin pool LV. If the input LV is being converted to
|
|
* a thin LV, a new lock is created for the thin pool and the lock from
|
|
* the input LV is freed. A cache pool LV has no lock, so the lock for
|
|
* the input LV is freed.
|
|
*/
|
|
if (vg_is_shared(vg)) {
|
|
lv->lock_args = NULL;
|
|
pool_lv->lock_args = NULL;
|
|
data_lv->lock_args = NULL;
|
|
metadata_lv->lock_args = NULL;
|
|
|
|
if (to_thin) {
|
|
if (!lockd_init_lv_args(cmd, vg, pool_lv, vg->lock_type, &pool_lv->lock_args)) {
|
|
log_error("Cannot allocate lock for new pool LV.");
|
|
goto bad;
|
|
}
|
|
} else if (to_thinpool) {
|
|
pool_lv->lock_args = lockd_data_args;
|
|
/* Don't free this lock below. */
|
|
lockd_data_args = NULL;
|
|
lockd_data_name = NULL;
|
|
}
|
|
|
|
/* Acquire the thin pool lock if the pool will remain active. */
|
|
if ((to_thin || to_thinpool) && is_active) {
|
|
if (!lockd_lv(cmd, pool_lv, "ex", LDLV_PERSISTENT)) {
|
|
log_error("Failed to lock new pool LV %s.", display_lvname(pool_lv));
|
|
goto bad;
|
|
}
|
|
lock_active_pool_done = 1;
|
|
}
|
|
}
|
|
|
|
if (to_thin) {
|
|
if (!lv_update_and_reload(lv))
|
|
goto_bad;
|
|
} else {
|
|
if (!vg_write(vg) || !vg_commit(vg))
|
|
goto_bad;
|
|
}
|
|
|
|
/*
|
|
* The main conversion is successfully committed. If any subsequent
|
|
* steps fail (creating spare, activating, unlocking), we do not
|
|
* currently have the ability to undo the changes committed up to this
|
|
* point. Failures in the remaining steps can print an error and cause
|
|
* the command to exit with an error, but no partial revert of the
|
|
* completed steps is attempted.
|
|
*/
|
|
log_print_unless_silent("Converted %s to %s %s.", converted_names,
|
|
(to_cachepool) ? "cache" : "thin",
|
|
(to_thin) ? "volume" : "pool");
|
|
|
|
if (activate_pool && !activate_lv(cmd, pool_lv)) {
|
|
log_error("Failed to activate pool logical volume %s.", display_lvname(pool_lv));
|
|
ret = 0;
|
|
}
|
|
|
|
/*
|
|
* Unlock and free locks that are no longer used.
|
|
*/
|
|
if (lockd_data_name) {
|
|
if (!lockd_lv_name(cmd, vg, lockd_data_name, &lockd_data_id, lockd_data_args, "un", lockd_data_flags)) {
|
|
log_error("Failed to unlock pool data LV %s/%s", vg->name, lockd_data_name);
|
|
ret = 0;
|
|
}
|
|
if (!lockd_free_lv(cmd, vg, lockd_data_name, &lockd_data_id, lockd_data_args)) {
|
|
log_error("Failed to free lock for pool data LV %s/%s", vg->name, lockd_data_name);
|
|
ret = 0;
|
|
}
|
|
}
|
|
if (lockd_meta_name) {
|
|
if (!lockd_lv_name(cmd, vg, lockd_meta_name, &lockd_meta_id, lockd_meta_args, "un", lockd_meta_flags)) {
|
|
log_error("Failed to unlock pool metadata LV %s/%s", vg->name, lockd_meta_name);
|
|
ret = 0;
|
|
}
|
|
if (!lockd_free_lv(cmd, vg, lockd_meta_name, &lockd_meta_id, lockd_meta_args)) {
|
|
log_error("Failed to free lock for pool metadata LV %s/%s", vg->name, lockd_meta_name);
|
|
ret = 0;
|
|
}
|
|
}
|
|
|
|
if (policy_settings)
|
|
dm_config_destroy(policy_settings);
|
|
|
|
if (!ret)
|
|
log_error("Manual intervention may be required to handle reported errors.");
|
|
|
|
return ret;
|
|
|
|
/*
|
|
* Error exit path for failures that occur before the main conversion
|
|
* is committed. Failures that occur after the main conversion is
|
|
* committed should not exit here. There is some cleanup missing here.
|
|
*/
|
|
bad:
|
|
if (vg_is_shared(vg)) {
|
|
if (lock_active_pool_done)
|
|
lockd_lv(cmd, pool_lv, "un", LDLV_PERSISTENT);
|
|
if (pool_lv && pool_lv->lock_args && pool_lv->new_lock_args)
|
|
lockd_free_lv(cmd, vg, pool_lv->name, &pool_lv->lvid.id[1], pool_lv->lock_args);
|
|
}
|
|
|
|
if (policy_settings)
|
|
dm_config_destroy(policy_settings);
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int _cache_vol_attach(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *lv_fast)
|
|
{
|
|
char cvol_name[NAME_LEN];
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *cache_lv;
|
|
uint32_t chunk_size = 0;
|
|
uint64_t poolmetadatasize = 0;
|
|
cache_metadata_format_t cache_metadata_format;
|
|
cache_mode_t cache_mode;
|
|
const char *policy_name;
|
|
struct dm_config_tree *policy_settings = NULL;
|
|
char *lockd_fast_args = NULL;
|
|
char *lockd_fast_name = NULL;
|
|
struct id lockd_fast_id;
|
|
int r = 0;
|
|
|
|
if (!validate_lv_cache_create_pool(lv_fast))
|
|
return_0;
|
|
|
|
if (!get_cache_params(cmd, &chunk_size, &cache_metadata_format, &cache_mode, &policy_name, &policy_settings))
|
|
goto_out;
|
|
|
|
/*
|
|
* lv/cache_lv keeps the same lockd lock it had before, the lock for
|
|
* lv_fast is kept but is not used while it's attached, and
|
|
* lv_corig has no lock. (When the cachevol is split a new lvmlockd
|
|
* lock does not need to be created for it again.)
|
|
*/
|
|
if (vg_is_shared(vg) && lv_fast->lock_args) {
|
|
lockd_fast_args = dm_pool_strdup(lv_fast->vg->vgmem, lv_fast->lock_args);
|
|
lockd_fast_name = dm_pool_strdup(lv_fast->vg->vgmem, lv_fast->name);
|
|
lockd_fast_id = lv_fast->lvid.id[1];
|
|
}
|
|
|
|
/*
|
|
* The lvm tradition is to rename an LV with a special role-specific
|
|
* suffix when it becomes hidden. Here the _cvol suffix is added to
|
|
* the fast LV name. When the cache is detached, it's renamed back.
|
|
*/
|
|
if (dm_snprintf(cvol_name, sizeof(cvol_name), "%s_cvol", lv_fast->name) < 0) {
|
|
log_error("Can't prepare new cachevol name for %s.", display_lvname(lv_fast));
|
|
goto out;
|
|
}
|
|
if (!lv_rename_update(cmd, lv_fast, cvol_name, 0))
|
|
goto_out;
|
|
|
|
lv_fast->status |= LV_CACHE_VOL; /* Mark as cachevol LV */
|
|
|
|
/*
|
|
* Changes the vg struct to match the desired state.
|
|
*
|
|
* - lv == cache_lv, which keeps existing lv name and id, gets new
|
|
* segment with segtype "cache".
|
|
*
|
|
* - lv_fast keeps its existing name and id, becomes hidden.
|
|
*
|
|
* - lv_corig gets new name (existing name + _corig suffix),
|
|
* gets new id, becomes hidden, gets segments from lv.
|
|
*/
|
|
|
|
if (!(cache_lv = lv_cache_create(lv_fast, lv)))
|
|
goto_out;
|
|
|
|
if (arg_is_set(cmd, poolmetadatasize_ARG))
|
|
poolmetadatasize = arg_uint64_value(cmd, poolmetadatasize_ARG, 0);
|
|
|
|
if (!cache_vol_set_params(cmd, cache_lv, lv_fast, poolmetadatasize, chunk_size, cache_metadata_format, cache_mode, policy_name, policy_settings))
|
|
goto_out;
|
|
|
|
if (cache_mode == CACHE_MODE_WRITEBACK) {
|
|
log_warn("WARNING: repairing a damaged cachevol is not yet possible.");
|
|
log_warn("WARNING: cache mode writethrough is suggested for safe operation.");
|
|
if (!arg_count(cmd, yes_ARG) &&
|
|
yes_no_prompt("Continue using writeback without repair?") == 'n')
|
|
goto_out;
|
|
}
|
|
|
|
/*
|
|
* vg_write(), suspend_lv(), vg_commit(), resume_lv(),
|
|
* where the old LV is suspended and the new LV is resumed.
|
|
*/
|
|
|
|
if (!lv_update_and_reload(cache_lv))
|
|
goto_out;
|
|
|
|
if (lockd_fast_name) {
|
|
/* lockd unlock for lv_fast */
|
|
if (!lockd_lv_name(cmd, vg, lockd_fast_name, &lockd_fast_id, lockd_fast_args, "un", LDLV_PERSISTENT))
|
|
log_error("Failed to unlock fast LV %s/%s", vg->name, lockd_fast_name);
|
|
}
|
|
|
|
r = 1;
|
|
out:
|
|
if (policy_settings)
|
|
dm_config_destroy(policy_settings);
|
|
|
|
return r;
|
|
}
|
|
|
|
static int _cache_pool_attach(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *cachepool_lv)
|
|
{
|
|
struct logical_volume *cache_lv;
|
|
uint32_t chunk_size = 0;
|
|
cache_metadata_format_t cache_metadata_format;
|
|
cache_mode_t cache_mode;
|
|
const char *policy_name;
|
|
struct dm_config_tree *policy_settings = NULL;
|
|
int r = 0;
|
|
|
|
if (!validate_lv_cache_create_pool(cachepool_lv))
|
|
return_0;
|
|
|
|
if (!get_cache_params(cmd, &chunk_size, &cache_metadata_format, &cache_mode, &policy_name, &policy_settings))
|
|
goto_bad;
|
|
|
|
if (!archive(lv->vg))
|
|
goto_bad;
|
|
|
|
if (!(cache_lv = lv_cache_create(cachepool_lv, lv)))
|
|
goto_bad;
|
|
|
|
if (!cache_set_params(first_seg(cache_lv), chunk_size, cache_metadata_format, cache_mode, policy_name, policy_settings))
|
|
goto_bad;
|
|
|
|
if (!lv_update_and_reload(cache_lv))
|
|
goto_bad;
|
|
|
|
r = 1;
|
|
bad:
|
|
if (policy_settings)
|
|
dm_config_destroy(policy_settings);
|
|
|
|
return r;
|
|
}
|
|
|
|
static struct convert_poll_id_list* _convert_poll_id_list_create(struct cmd_context *cmd,
|
|
const struct logical_volume *lv)
|
|
{
|
|
struct convert_poll_id_list *idl = (struct convert_poll_id_list *) dm_pool_alloc(cmd->mem, sizeof(struct convert_poll_id_list));
|
|
|
|
if (!idl) {
|
|
log_error("Convert poll ID list allocation failed.");
|
|
return NULL;
|
|
}
|
|
|
|
if (!(idl->id = _create_id(cmd, lv->vg->name, lv->name, lv->lvid.s))) {
|
|
dm_pool_free(cmd->mem, idl);
|
|
return_NULL;
|
|
}
|
|
|
|
idl->is_merging_origin = lv_is_merging_origin(lv);
|
|
idl->is_merging_origin_thin = idl->is_merging_origin && seg_is_thin_volume(find_snapshot(lv));
|
|
|
|
return idl;
|
|
}
|
|
|
|
/*
|
|
* Data/results accumulated during processing.
|
|
*/
|
|
struct lvconvert_result {
|
|
unsigned need_polling:1;
|
|
unsigned wait_cleaner_writecache:1;
|
|
unsigned active_begin:1;
|
|
unsigned remove_cache:1;
|
|
struct dm_list poll_idls;
|
|
};
|
|
|
|
|
|
/*
|
|
* repair-related lvconvert utilities
|
|
*/
|
|
|
|
static int _lvconvert_repair_pvs_mirror(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle,
|
|
struct dm_list *use_pvh)
|
|
{
|
|
struct lvconvert_result *lr = (struct lvconvert_result *) handle->custom_handle;
|
|
struct lvconvert_params lp = { 0 };
|
|
struct convert_poll_id_list *idl;
|
|
int ret;
|
|
|
|
/*
|
|
* We want to allow cmirror active on multiple nodes to be repaired,
|
|
* but normal mirror to only be repaired if active exclusively here.
|
|
* If the LV is active it already has the necessary lock, but if not
|
|
* active, then require ex since we cannot know the active state on
|
|
* other hosts.
|
|
*/
|
|
if (!lv_is_active(lv)) {
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
return_0;
|
|
}
|
|
|
|
/*
|
|
* FIXME: temporary use of lp because _lvconvert_mirrors_repair()
|
|
* and _aux() still use lp fields everywhere.
|
|
* Migrate them away from using lp (for the most part just use
|
|
* local variables, and check arg_values directly).
|
|
*/
|
|
|
|
/*
|
|
* Fill in any lp fields here that this fn expects to be set before
|
|
* it's called. It's hard to tell what the old code expects in lp
|
|
* for repair; it doesn't take the stripes option, but it seems to
|
|
* expect lp.stripes to be set to 1.
|
|
*/
|
|
lp.alloc = (alloc_policy_t) arg_uint_value(cmd, alloc_ARG, ALLOC_INHERIT);
|
|
lp.stripes = 1;
|
|
|
|
ret = _lvconvert_mirrors_repair(cmd, lv, &lp, use_pvh);
|
|
|
|
if (lp.need_polling) {
|
|
if (!lv_is_active(lv))
|
|
log_print_unless_silent("Conversion starts after activation.");
|
|
else {
|
|
if (!(idl = _convert_poll_id_list_create(cmd, lv)))
|
|
return 0;
|
|
dm_list_add(&lr->poll_idls, &idl->list);
|
|
}
|
|
lr->need_polling = 1;
|
|
}
|
|
|
|
return ret;
|
|
}
|
|
|
|
static void _lvconvert_repair_pvs_raid_ask(struct cmd_context *cmd, int *do_it)
|
|
{
|
|
const char *dev_policy;
|
|
|
|
*do_it = 1;
|
|
|
|
if (arg_is_set(cmd, usepolicies_ARG)) {
|
|
dev_policy = find_config_tree_str(cmd, activation_raid_fault_policy_CFG, NULL);
|
|
|
|
if (!strcmp(dev_policy, "allocate") ||
|
|
!strcmp(dev_policy, "replace"))
|
|
return;
|
|
|
|
/* else if (!strcmp(dev_policy, "anything_else")) -- no replace */
|
|
*do_it = 0;
|
|
return;
|
|
}
|
|
|
|
if (!arg_count(cmd, yes_ARG) &&
|
|
yes_no_prompt("Attempt to replace failed RAID images "
|
|
"(requires full device resync)? [y/n]: ") == 'n') {
|
|
*do_it = 0;
|
|
}
|
|
}
|
|
|
|
static int _lvconvert_repair_pvs_raid(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle,
|
|
struct dm_list *use_pvh)
|
|
{
|
|
struct dm_list *failed_pvs;
|
|
int do_it;
|
|
|
|
if (!lv_is_active(lv_lock_holder(lv))) {
|
|
log_error("%s must be active to perform this operation.", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
lv_check_transient(lv); /* TODO check this in lib for all commands? */
|
|
|
|
_lvconvert_repair_pvs_raid_ask(cmd, &do_it);
|
|
|
|
if (do_it) {
|
|
if (!(failed_pvs = _failed_pv_list(lv->vg)))
|
|
return_0;
|
|
|
|
if (!lv_raid_replace(lv, arg_count(cmd, force_ARG), failed_pvs, use_pvh)) {
|
|
log_error("Failed to replace faulty devices in %s.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
log_print_unless_silent("Faulty devices in %s successfully replaced.",
|
|
display_lvname(lv));
|
|
return 1;
|
|
}
|
|
|
|
/* "warn" if policy not set to replace */
|
|
if (arg_is_set(cmd, usepolicies_ARG))
|
|
log_warn("Use 'lvconvert --repair %s' to replace "
|
|
"failed device.", display_lvname(lv));
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_repair_pvs(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct dm_list *failed_pvs;
|
|
struct dm_list *use_pvh;
|
|
int ret;
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
|
|
return_ECMD_FAILED;
|
|
} else
|
|
use_pvh = &lv->vg->pvs;
|
|
|
|
if (lv_is_raid(lv))
|
|
ret = _lvconvert_repair_pvs_raid(cmd, lv, handle, use_pvh);
|
|
else if (lv_is_mirror(lv))
|
|
ret = _lvconvert_repair_pvs_mirror(cmd, lv, handle, use_pvh);
|
|
else
|
|
ret = 0;
|
|
|
|
if (ret && arg_is_set(cmd, usepolicies_ARG)) {
|
|
if ((failed_pvs = _failed_pv_list(lv->vg)))
|
|
_remove_missing_empty_pv(lv->vg, failed_pvs);
|
|
}
|
|
|
|
return ret ? ECMD_PROCESSED : ECMD_FAILED;
|
|
}
|
|
|
|
static int _lvconvert_repair_cachepool_thinpool(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
int poolmetadataspare = arg_int_value(cmd, poolmetadataspare_ARG, DEFAULT_POOL_METADATA_SPARE);
|
|
struct dm_list *use_pvh;
|
|
|
|
/* ensure it's not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
return_0;
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
|
|
return_ECMD_FAILED;
|
|
} else
|
|
use_pvh = &lv->vg->pvs;
|
|
|
|
if (lv_is_thin_pool(lv)) {
|
|
if (!_lvconvert_thin_pool_repair(cmd, lv, use_pvh, poolmetadataspare))
|
|
return_ECMD_FAILED;
|
|
} else /* cache */ {
|
|
if (!_lvconvert_cache_repair(cmd, lv, use_pvh, poolmetadataspare))
|
|
return_ECMD_FAILED;
|
|
}
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
static int _lvconvert_repair_single(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
if (lv_is_thin_pool(lv) ||
|
|
lv_is_cache(lv) ||
|
|
lv_is_cache_pool(lv))
|
|
return _lvconvert_repair_cachepool_thinpool(cmd, lv, handle);
|
|
|
|
if (lv_is_raid(lv) || lv_is_mirror(lv))
|
|
return _lvconvert_repair_pvs(cmd, lv, handle);
|
|
|
|
log_error("Unsupported volume type for repair of volume %s.",
|
|
display_lvname(lv));
|
|
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
/*
|
|
* FIXME: add option --repair-pvs to call _lvconvert_repair_pvs() directly,
|
|
* and option --repair-thinpool to call _lvconvert_repair_thinpool().
|
|
* and option --repair-cache to call _lvconvert_repair_cache().
|
|
* and option --repair-cachepool to call _lvconvert_repair_cachepool().
|
|
*/
|
|
int lvconvert_repair_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_result lr = { 0 };
|
|
struct convert_poll_id_list *idl;
|
|
int saved_ignore_suspended_devices;
|
|
int ret, poll_ret;
|
|
|
|
dm_list_init(&lr.poll_idls);
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lr;
|
|
|
|
saved_ignore_suspended_devices = ignore_suspended_devices();
|
|
init_ignore_suspended_devices(1);
|
|
|
|
cmd->handles_missing_pvs = 1;
|
|
|
|
ret = process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
handle, NULL, &_lvconvert_repair_single);
|
|
|
|
init_ignore_suspended_devices(saved_ignore_suspended_devices);
|
|
|
|
if (lr.need_polling) {
|
|
dm_list_iterate_items(idl, &lr.poll_idls) {
|
|
poll_ret = _lvconvert_poll_by_id(cmd, idl->id,
|
|
arg_is_set(cmd, background_ARG), 0, 0);
|
|
if (poll_ret > ret)
|
|
ret = poll_ret;
|
|
}
|
|
}
|
|
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
static int _lvconvert_replace_pv_single(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct arg_value_group_list *group;
|
|
const char *tmp_str;
|
|
struct dm_list *use_pvh;
|
|
struct dm_list *replace_pvh;
|
|
char **replace_pvs;
|
|
int replace_pv_count;
|
|
int i;
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
|
|
return_ECMD_FAILED;
|
|
} else
|
|
use_pvh = &lv->vg->pvs;
|
|
|
|
if (!(replace_pv_count = arg_count(cmd, replace_ARG)))
|
|
return_ECMD_FAILED;
|
|
|
|
if (!(replace_pvs = dm_pool_alloc(cmd->mem, sizeof(char *) * replace_pv_count)))
|
|
return_ECMD_FAILED;
|
|
|
|
i = 0;
|
|
dm_list_iterate_items(group, &cmd->arg_value_groups) {
|
|
if (!grouped_arg_is_set(group->arg_values, replace_ARG))
|
|
continue;
|
|
if (!(tmp_str = grouped_arg_str_value(group->arg_values, replace_ARG, NULL))) {
|
|
log_error("Failed to get '--replace' argument");
|
|
return ECMD_FAILED;
|
|
}
|
|
if (!(replace_pvs[i++] = dm_pool_strdup(cmd->mem, tmp_str)))
|
|
return_ECMD_FAILED;
|
|
}
|
|
|
|
if (!(replace_pvh = create_pv_list(cmd->mem, lv->vg, replace_pv_count, replace_pvs, 0)))
|
|
return_ECMD_FAILED;
|
|
|
|
if (!lv_raid_replace(lv, arg_count(cmd, force_ARG), replace_pvh, use_pvh))
|
|
return_ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_replace_pv_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_result lr = { 0 };
|
|
int ret;
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lr;
|
|
|
|
ret = process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
handle, NULL, &_lvconvert_replace_pv_single);
|
|
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
|
|
/*
|
|
* Merge a COW snapshot LV into its origin.
|
|
*/
|
|
|
|
static int _lvconvert_merge_snapshot_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct lvconvert_result *lr = (struct lvconvert_result *) handle->custom_handle;
|
|
struct logical_volume *lv_to_poll = NULL;
|
|
struct convert_poll_id_list *idl;
|
|
|
|
if (!_lvconvert_merge_old_snapshot(cmd, lv, &lv_to_poll))
|
|
return_ECMD_FAILED;
|
|
|
|
if (lv_to_poll) {
|
|
if (!(idl = _convert_poll_id_list_create(cmd, lv_to_poll)))
|
|
return_ECMD_FAILED;
|
|
dm_list_add(&lr->poll_idls, &idl->list);
|
|
lr->need_polling = 1;
|
|
}
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_merge_snapshot_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_result lr = { 0 };
|
|
struct convert_poll_id_list *idl;
|
|
int ret, poll_ret;
|
|
|
|
dm_list_init(&lr.poll_idls);
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lr;
|
|
|
|
ret = process_each_lv(cmd, cmd->position_argc, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
handle, NULL, &_lvconvert_merge_snapshot_single);
|
|
|
|
if (lr.need_polling) {
|
|
dm_list_iterate_items(idl, &lr.poll_idls) {
|
|
poll_ret = _lvconvert_poll_by_id(cmd, idl->id,
|
|
arg_is_set(cmd, background_ARG), 1, 0);
|
|
if (poll_ret > ret)
|
|
ret = poll_ret;
|
|
}
|
|
}
|
|
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
/*
|
|
* Separate a COW snapshot from its origin.
|
|
*
|
|
* lvconvert --splitsnapshot LV_snapshot
|
|
* lvconvert_split_cow_snapshot
|
|
*/
|
|
|
|
static int _lvconvert_split_snapshot_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
if (!_lvconvert_splitsnapshot(cmd, lv))
|
|
return_ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_split_snapshot_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_split_snapshot_single);
|
|
}
|
|
|
|
/*
|
|
* Combine two LVs that were once an origin/cow pair of LVs, were then
|
|
* separated with --splitsnapshot, and now with this command are combined again
|
|
* into the origin/cow pair.
|
|
*
|
|
* This is an obscure command that has little to no real uses.
|
|
*
|
|
* The command has unusual handling of position args. The first position arg
|
|
* will become the origin LV, and is not processed by process_each_lv. The
|
|
* second position arg will become the cow LV and is processed by
|
|
* process_each_lv.
|
|
*
|
|
* The single function can grab the origin LV from position_argv[0].
|
|
*
|
|
* begin with an ordinary LV foo:
|
|
* lvcreate -n foo -L 1 vg
|
|
*
|
|
* create a cow snapshot of foo named foosnap:
|
|
* lvcreate -s -L 1 -n foosnap vg/foo
|
|
*
|
|
* now, foo is an "origin LV" and foosnap is a "cow LV"
|
|
* (foosnap matches LV_snapshot aka lv_is_cow)
|
|
*
|
|
* split the two LVs apart:
|
|
* lvconvert --splitsnapshot vg/foosnap
|
|
*
|
|
* now, foo is *not* an origin LV and foosnap is *not* a cow LV
|
|
* (foosnap does not match LV_snapshot)
|
|
*
|
|
* now, combine the two LVs again:
|
|
* lvconvert --snapshot vg/foo vg/foosnap
|
|
*
|
|
* after this, foosnap will match LV_snapshot again.
|
|
*
|
|
* FIXME: when splitsnapshot is run, the previous cow LV should be
|
|
* flagged in the metadata somehow, and then that flag should be
|
|
* required here. As it is now, the first and second args
|
|
* (origin and cow) can be swapped and nothing catches it.
|
|
*/
|
|
|
|
static int _lvconvert_combine_split_snapshot_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
const char *origin_name = cmd->position_argv[0];
|
|
|
|
if (vg_is_shared(lv->vg)) {
|
|
log_error("Unable to combine split snapshots in VG with lock_type %s", lv->vg->lock_type);
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
/* If origin_name includes VG name, the VG name is removed. */
|
|
if (!validate_lvname_param(cmd, &lv->vg->name, &origin_name))
|
|
return_ECMD_FAILED;
|
|
|
|
if (!_lvconvert_snapshot(cmd, lv, origin_name))
|
|
return_ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_combine_split_snapshot_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
const char *vgname = NULL;
|
|
const char *lvname1_orig;
|
|
const char *lvname2_orig;
|
|
const char *lvname1_split;
|
|
char *vglv;
|
|
int vglv_sz;
|
|
|
|
/*
|
|
* Hack to accomodate an old parsing quirk that allowed the
|
|
* the VG name to be attached to only the LV in arg pos 1,
|
|
* i.e. lvconvert -s vgname/lvname lvname
|
|
*
|
|
* The LV name in arg pos 2 is the one that is processed
|
|
* by process_each_lv(). If that LV has no VG name, but
|
|
* the first LV does, then copy the VG name from arg pos 1
|
|
* and add it to the LV name in arg pos 2 so that the
|
|
* standard arg parsing in process_each_lv will find it.
|
|
*
|
|
* This is the only instance in all commands.
|
|
*/
|
|
|
|
lvname1_orig = cmd->position_argv[0];
|
|
lvname2_orig = cmd->position_argv[1];
|
|
|
|
if (strchr(lvname1_orig, '/') && !strchr(lvname2_orig, '/') && !getenv("LVM_VG_NAME")) {
|
|
if (!(lvname1_split = dm_pool_strdup(cmd->mem, lvname1_orig)))
|
|
return_ECMD_FAILED;
|
|
|
|
if (!validate_lvname_param(cmd, &vgname, &lvname1_split))
|
|
return_ECMD_FAILED;
|
|
|
|
vglv_sz = strlen(vgname) + strlen(lvname2_orig) + 2;
|
|
if (!(vglv = dm_pool_alloc(cmd->mem, vglv_sz)) ||
|
|
dm_snprintf(vglv, vglv_sz, "%s/%s", vgname, lvname2_orig) < 0) {
|
|
log_error("vg/lv string alloc failed.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
/* vglv is now vgname/lvname2 and replaces lvname2_orig */
|
|
|
|
cmd->position_argv[1] = vglv;
|
|
}
|
|
|
|
return process_each_lv(cmd, 1, cmd->position_argv + 1, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_combine_split_snapshot_single);
|
|
}
|
|
|
|
static int _lvconvert_start_poll_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct lvconvert_result *lr = (struct lvconvert_result *) handle->custom_handle;
|
|
struct convert_poll_id_list *idl;
|
|
|
|
if (!(idl = _convert_poll_id_list_create(cmd, lv)))
|
|
return_ECMD_FAILED;
|
|
dm_list_add(&lr->poll_idls, &idl->list);
|
|
|
|
lr->need_polling = 1;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_start_poll_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_result lr = { 0 };
|
|
struct convert_poll_id_list *idl;
|
|
int saved_ignore_suspended_devices;
|
|
int ret, poll_ret;
|
|
|
|
dm_list_init(&lr.poll_idls);
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lr;
|
|
|
|
saved_ignore_suspended_devices = ignore_suspended_devices();
|
|
init_ignore_suspended_devices(1);
|
|
|
|
cmd->handles_missing_pvs = 1;
|
|
|
|
ret = process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
handle, NULL, &_lvconvert_start_poll_single);
|
|
|
|
init_ignore_suspended_devices(saved_ignore_suspended_devices);
|
|
|
|
if (lr.need_polling) {
|
|
dm_list_iterate_items(idl, &lr.poll_idls) {
|
|
poll_ret = _lvconvert_poll_by_id(cmd, idl->id,
|
|
arg_is_set(cmd, background_ARG), 0, 0);
|
|
if (poll_ret > ret)
|
|
ret = poll_ret;
|
|
}
|
|
}
|
|
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
static int _lvconvert_to_pool_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct dm_list *use_pvh = NULL;
|
|
int to_thinpool = 0;
|
|
int to_cachepool = 0;
|
|
|
|
switch (cmd->command->command_enum) {
|
|
case lvconvert_to_thinpool_CMD:
|
|
to_thinpool = 1;
|
|
break;
|
|
case lvconvert_to_cachepool_CMD:
|
|
to_cachepool = 1;
|
|
break;
|
|
default:
|
|
log_error(INTERNAL_ERROR "Invalid lvconvert pool command");
|
|
return ECMD_FAILED;
|
|
};
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
|
|
return_ECMD_FAILED;
|
|
} else
|
|
use_pvh = &lv->vg->pvs;
|
|
|
|
if (!_lvconvert_to_pool(cmd, lv, lv, to_thinpool, to_cachepool, 0, use_pvh))
|
|
return_ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
/*
|
|
* The LV position arg is used as thinpool/cachepool data LV.
|
|
*/
|
|
|
|
int lvconvert_to_pool_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_to_pool_single);
|
|
}
|
|
|
|
#define MAX_CACHEDEVS 8
|
|
|
|
static int _lv_create_cachevol(struct cmd_context *cmd,
|
|
struct volume_group *vg,
|
|
struct logical_volume *lv,
|
|
struct logical_volume **cachevol_lv)
|
|
{
|
|
char cvname[NAME_LEN];
|
|
char format[NAME_LEN];
|
|
struct dm_list *use_pvh;
|
|
struct pv_list *pvl;
|
|
const char *device_name = "";
|
|
struct device *dev_fast;
|
|
char *dev_argv[MAX_CACHEDEVS];
|
|
int dev_argc = 0;
|
|
uint64_t cache_size_sectors = 0;
|
|
uint64_t full_size_sectors = 0;
|
|
uint64_t pv_size_sectors;
|
|
struct logical_volume *cachevol;
|
|
struct arg_value_group_list *group;
|
|
struct lvcreate_params lp = {
|
|
.activate = CHANGE_AN,
|
|
.alloc = ALLOC_INHERIT,
|
|
.major = -1,
|
|
.minor = -1,
|
|
.permission = LVM_READ | LVM_WRITE,
|
|
.pvh = &vg->pvs,
|
|
.read_ahead = DM_READ_AHEAD_NONE,
|
|
.stripes = 1,
|
|
.vg_name = vg->name,
|
|
.zero = 0,
|
|
.wipe_signatures = 0,
|
|
.suppress_zero_warn = 1,
|
|
};
|
|
|
|
/*
|
|
* If cache size is not set, and all cachedevice's are unused,
|
|
* then the cache size is the sum of all cachedevice sizes.
|
|
*/
|
|
cache_size_sectors = arg_uint64_value(cmd, cachesize_ARG, 0);
|
|
|
|
dm_list_iterate_items(group, &cmd->arg_value_groups) {
|
|
if (!grouped_arg_is_set(group->arg_values, cachedevice_ARG))
|
|
continue;
|
|
|
|
if (!(device_name = grouped_arg_str_value(group->arg_values, cachedevice_ARG, NULL)))
|
|
break;
|
|
|
|
if (device_name[0] == '@') {
|
|
if (!cache_size_sectors) {
|
|
log_error("With tag as cachedevice, --cachesize is required.");
|
|
return 0;
|
|
}
|
|
goto add_dev_arg;
|
|
}
|
|
|
|
if (!(dev_fast = dev_cache_get(cmd, device_name, cmd->filter))) {
|
|
log_error("Device %s not found.", device_name);
|
|
return 0;
|
|
}
|
|
|
|
if (!(pvl = find_pv_in_vg(vg, device_name))) {
|
|
log_error("PV %s not found in VG.", device_name);
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* If the dev is used in the VG, then require a cachesize to allocate
|
|
* from it. If it is not used in the VG, then prompt asking if the
|
|
* entire dev should be used.
|
|
*/
|
|
if (!cache_size_sectors && pvl->pv->pe_alloc_count) {
|
|
log_error("PV %s is in use, --cachesize is required.", device_name);
|
|
return 0;
|
|
}
|
|
|
|
if (!cache_size_sectors) {
|
|
pv_size_sectors = (pvl->pv->pe_count * (uint64_t)vg->extent_size);
|
|
|
|
if (!arg_is_set(cmd, yes_ARG) &&
|
|
yes_no_prompt("Use all %s from %s for cache? [y/n]: ",
|
|
display_size(cmd, pv_size_sectors), device_name) == 'n') {
|
|
log_print_unless_silent("Use --cachesize SizeMB to use a part of the cachedevice.");
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
full_size_sectors += pv_size_sectors;
|
|
}
|
|
add_dev_arg:
|
|
if (dev_argc >= MAX_CACHEDEVS) {
|
|
log_error("Cannot allocate from more than %u cache devices.", MAX_CACHEDEVS);
|
|
return 0;
|
|
}
|
|
|
|
dev_argv[dev_argc++] = (char*)device_name;
|
|
}
|
|
|
|
if (!cache_size_sectors)
|
|
cache_size_sectors = full_size_sectors;
|
|
|
|
if (!dev_argc) {
|
|
log_error("No cachedevice specified to create a cachevol.");
|
|
return 0;
|
|
}
|
|
|
|
if (!(use_pvh = create_pv_list(cmd->mem, vg, dev_argc, dev_argv, 1))) {
|
|
log_error("cachedevice not found in VG %s.", device_name);
|
|
return 0;
|
|
}
|
|
|
|
if (dm_snprintf(cvname, NAME_LEN, "%s_cache", lv->name) < 0) {
|
|
log_error("Failed to create cachevol LV name.");
|
|
return 0;
|
|
}
|
|
|
|
if (find_lv(vg, cvname)) {
|
|
memset(format, 0, sizeof(cvname));
|
|
memset(cvname, 0, sizeof(cvname));
|
|
if (dm_snprintf(format, sizeof(format), "%s_cache%%d", lv->name) < 0) {
|
|
log_error("Failed to generate cachevol LV format.");
|
|
return 0;
|
|
}
|
|
if (!generate_lv_name(vg, format, cvname, sizeof(cvname))) {
|
|
log_error("Failed to generate cachevol LV name.");
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
lp.lv_name = cvname;
|
|
lp.pvh = use_pvh;
|
|
lp.extents = cache_size_sectors / vg->extent_size;
|
|
|
|
log_print_unless_silent("Creating cachevol LV %s with size %s.",
|
|
cvname, display_size(cmd, cache_size_sectors));
|
|
|
|
dm_list_init(&lp.tags);
|
|
|
|
if (!(lp.segtype = get_segtype_from_string(cmd, SEG_TYPE_NAME_STRIPED)))
|
|
return_0;
|
|
|
|
if (!(cachevol = lv_create_single(vg, &lp))) {
|
|
log_error("Failed to create cachevol LV");
|
|
return 0;
|
|
}
|
|
|
|
*cachevol_lv = cachevol;
|
|
return 1;
|
|
}
|
|
|
|
int lvconvert_cachevol_attach_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *lv_fast;
|
|
const char *fast_name;
|
|
|
|
/*
|
|
* User specifies an existing cachevol to use or a cachedevice
|
|
* to create a cachevol from.
|
|
*/
|
|
if ((fast_name = arg_str_value(cmd, cachevol_ARG, NULL))) {
|
|
if (!validate_lvname_param(cmd, &vg->name, &fast_name))
|
|
goto_bad;
|
|
|
|
if (!(lv_fast = find_lv(vg, fast_name))) {
|
|
log_error("LV %s not found.", fast_name);
|
|
goto bad;
|
|
}
|
|
|
|
if (lv_is_cache_vol(lv_fast)) {
|
|
log_error("LV %s is already used as a cachevol.", display_lvname(lv_fast));
|
|
goto bad;
|
|
}
|
|
|
|
if (!dm_list_empty(&lv_fast->segs_using_this_lv)) {
|
|
log_error("LV %s is already in use.", display_lvname(lv_fast));
|
|
goto bad;
|
|
}
|
|
|
|
if (!arg_is_set(cmd, yes_ARG) &&
|
|
yes_no_prompt("Erase all existing data on %s? [y/n]: ", display_lvname(lv_fast)) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
goto bad;
|
|
}
|
|
|
|
if (!lockd_lv(cmd, lv_fast, "ex", 0))
|
|
goto_bad;
|
|
} else {
|
|
if (!_lv_create_cachevol(cmd, vg, lv, &lv_fast))
|
|
goto_bad;
|
|
}
|
|
|
|
/* Ensure the LV is not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
goto_bad;
|
|
|
|
if (!wipe_cache_pool(lv_fast))
|
|
goto_bad;
|
|
|
|
/* When the lv arg is a thinpool, redirect command to data sub lv. */
|
|
|
|
if (lv_is_thin_pool(lv)) {
|
|
lv = seg_lv(first_seg(lv), 0);
|
|
log_verbose("Redirecting operation to data sub LV %s.", display_lvname(lv));
|
|
}
|
|
|
|
if (!_raid_split_image_conversion(lv))
|
|
goto_bad;
|
|
|
|
/* Attach the cache to the main LV. */
|
|
|
|
if (!_cache_vol_attach(cmd, lv, lv_fast))
|
|
goto_bad;
|
|
|
|
log_print_unless_silent("Logical volume %s is now cached.", display_lvname(lv));
|
|
|
|
return ECMD_PROCESSED;
|
|
bad:
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
static int _lvconvert_cachepool_attach_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *cachepool_lv;
|
|
const char *cachepool_name;
|
|
|
|
if (!(cachepool_name = arg_str_value(cmd, cachepool_ARG, NULL)))
|
|
goto_out;
|
|
|
|
if (!validate_lvname_param(cmd, &vg->name, &cachepool_name))
|
|
goto_out;
|
|
|
|
if (!(cachepool_lv = find_lv(vg, cachepool_name))) {
|
|
log_error("Cache pool %s not found.", cachepool_name);
|
|
goto out;
|
|
}
|
|
|
|
if (!validate_lv_cache_create_origin(lv))
|
|
goto_out;
|
|
|
|
/* Ensure the LV is not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
goto_out;
|
|
|
|
/*
|
|
* If cachepool_lv is not yet a cache pool, convert it to one.
|
|
* If using an existing cache pool, wipe it.
|
|
*/
|
|
|
|
if (!lv_is_cache_pool(cachepool_lv)) {
|
|
int lvt_enum = get_lvt_enum(cachepool_lv);
|
|
const struct lv_type *lvtype = get_lv_type(lvt_enum);
|
|
|
|
if (lvt_enum != striped_LVT && lvt_enum != linear_LVT && lvt_enum != raid_LVT) {
|
|
log_error("LV %s with type %s cannot be converted to a cache pool.",
|
|
display_lvname(cachepool_lv), lvtype ? lvtype->name : "unknown");
|
|
goto out;
|
|
}
|
|
|
|
if (lv_is_cache_vol(cachepool_lv)) {
|
|
log_error("LV %s is already used as a cachevol.", display_lvname(cachepool_lv));
|
|
goto out;
|
|
}
|
|
|
|
if (cachepool_lv == lv) {
|
|
log_error("Use a different LV for cache pool LV and cache LV %s.",
|
|
display_lvname(cachepool_lv));
|
|
goto out;
|
|
}
|
|
|
|
if (!_lvconvert_to_pool(cmd, cachepool_lv, lv, 0, 1, 0, &vg->pvs)) {
|
|
log_error("LV %s could not be converted to a cache pool.",
|
|
display_lvname(cachepool_lv));
|
|
goto out;
|
|
}
|
|
} else {
|
|
if (!dm_list_empty(&cachepool_lv->segs_using_this_lv)) {
|
|
log_error("Cache pool %s is already in use.", cachepool_name);
|
|
goto out;
|
|
}
|
|
|
|
/* Note: requires rather deep know-how to skip zeroing */
|
|
if (!arg_is_set(cmd, zero_ARG)) {
|
|
if (!arg_is_set(cmd, yes_ARG) &&
|
|
yes_no_prompt("Do you want wipe existing metadata of cache pool %s? [y/n]: ",
|
|
display_lvname(cachepool_lv)) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
log_error("To preserve cache metadata add option \"--zero n\".");
|
|
log_warn("WARNING: Reusing mismatched cache pool metadata MAY DESTROY YOUR DATA!");
|
|
goto out;
|
|
}
|
|
/* Wiping confirmed, go ahead */
|
|
if (!wipe_cache_pool(cachepool_lv))
|
|
goto_out;
|
|
} else if (arg_int_value(cmd, zero_ARG, 0)) {
|
|
if (!wipe_cache_pool(cachepool_lv))
|
|
goto_out;
|
|
} else {
|
|
log_warn("WARNING: Reusing cache pool metadata %s for volume caching.",
|
|
display_lvname(cachepool_lv));
|
|
}
|
|
}
|
|
|
|
/* When the lv arg is a thinpool, redirect command to data sub lv. */
|
|
|
|
if (lv_is_thin_pool(lv)) {
|
|
lv = seg_lv(first_seg(lv), 0);
|
|
log_verbose("Redirecting operation to data sub LV %s.", display_lvname(lv));
|
|
} else if (lv_is_vdo_pool(lv)) {
|
|
lv = seg_lv(first_seg(lv), 0);
|
|
log_verbose("Redirecting operation to data sub LV %s.", display_lvname(lv));
|
|
}
|
|
|
|
if (!_raid_split_image_conversion(lv))
|
|
goto_out;
|
|
|
|
/* Attach the cache to the main LV. */
|
|
|
|
if (!_cache_pool_attach(cmd, lv, cachepool_lv))
|
|
goto_out;
|
|
|
|
log_print_unless_silent("Logical volume %s is now cached.", display_lvname(lv));
|
|
|
|
return ECMD_PROCESSED;
|
|
out:
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
int lvconvert_to_cache_with_cachepool_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_cachepool_attach_single);
|
|
}
|
|
|
|
static int _lvconvert_to_thin_with_external_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *thinpool_lv;
|
|
const char *thinpool_name;
|
|
|
|
if (!(thinpool_name = arg_str_value(cmd, thinpool_ARG, NULL)))
|
|
goto_out;
|
|
|
|
if (!validate_lvname_param(cmd, &vg->name, &thinpool_name))
|
|
goto_out;
|
|
|
|
if (!(thinpool_lv = find_lv(vg, thinpool_name))) {
|
|
log_error("Thin pool %s not found.", thinpool_name);
|
|
goto out;
|
|
}
|
|
|
|
/* If thinpool_lv is not yet a thin pool, convert it to one. */
|
|
|
|
if (!lv_is_thin_pool(thinpool_lv)) {
|
|
int lvt_enum = get_lvt_enum(thinpool_lv);
|
|
const struct lv_type *lvtype = get_lv_type(lvt_enum);
|
|
|
|
if (lvt_enum != striped_LVT && lvt_enum != linear_LVT && lvt_enum != raid_LVT) {
|
|
log_error("LV %s with type %s cannot be converted to a thin pool.",
|
|
display_lvname(thinpool_lv), lvtype ? lvtype->name : "unknown");
|
|
goto out;
|
|
}
|
|
|
|
if (thinpool_lv == lv) {
|
|
log_error("Use a different LV for thin pool LV and thin LV %s.",
|
|
display_lvname(thinpool_lv));
|
|
goto out;
|
|
}
|
|
|
|
if (!_lvconvert_to_pool(cmd, thinpool_lv, lv, 1, 0, 0, &vg->pvs)) {
|
|
log_error("LV %s could not be converted to a thin pool.",
|
|
display_lvname(thinpool_lv));
|
|
goto out;
|
|
}
|
|
|
|
if (!(thinpool_lv = find_lv(vg, thinpool_name))) {
|
|
log_error(INTERNAL_ERROR "LV %s cannot be found.", thinpool_name);
|
|
goto out;
|
|
}
|
|
|
|
if (!lv_is_thin_pool(thinpool_lv)) {
|
|
log_error(INTERNAL_ERROR "LV %s is not a thin pool.", display_lvname(thinpool_lv));
|
|
goto out;
|
|
}
|
|
}
|
|
|
|
/* If lv is a cache volume, all data must be flushed. */
|
|
|
|
if (lv_is_cache(lv)) {
|
|
const struct lv_segment *pool_seg = first_seg(first_seg(lv)->pool_lv);
|
|
int is_clean;
|
|
|
|
if (pool_seg->cache_mode != CACHE_MODE_WRITETHROUGH) {
|
|
log_error("Cannot convert cache volume %s with %s cache mode to external origin.",
|
|
display_lvname(lv), get_cache_mode_name(pool_seg));
|
|
log_error("To proceed, run 'lvchange --cachemode writethrough %s'.",
|
|
display_lvname(lv));
|
|
goto out;
|
|
}
|
|
|
|
if (!lv_cache_wait_for_clean(lv, &is_clean))
|
|
goto_out;
|
|
|
|
if (!is_clean) {
|
|
log_error("Cache %s is not clean, refusing to convert to external origin.",
|
|
display_lvname(lv));
|
|
goto out;
|
|
}
|
|
}
|
|
|
|
/* Convert lv to thin with external origin using thinpool_lv. */
|
|
|
|
if (!_lvconvert_to_thin_with_external(cmd, lv, thinpool_lv))
|
|
goto_out;
|
|
|
|
return ECMD_PROCESSED;
|
|
|
|
out:
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
int lvconvert_to_thin_with_external_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_to_thin_with_external_single);
|
|
}
|
|
|
|
static int _lvconvert_to_thin_with_data(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct dm_list *use_pvh;
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1,
|
|
cmd->position_argv + 1, 0)))
|
|
return_ECMD_FAILED;
|
|
} else
|
|
use_pvh = &lv->vg->pvs;
|
|
|
|
if (!_lvconvert_to_pool(cmd, lv, lv, 1, 0, 1, use_pvh)) {
|
|
log_error("LV %s could not be converted to a thin volume.",
|
|
display_lvname(lv));
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_to_thin_with_data_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_to_thin_with_data);
|
|
}
|
|
|
|
static int _lvconvert_swap_pool_metadata_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *metadata_lv;
|
|
const char *metadata_name;
|
|
|
|
if (!(metadata_name = arg_str_value(cmd, poolmetadata_ARG, NULL)))
|
|
goto_out;
|
|
|
|
if (!validate_lvname_param(cmd, &vg->name, &metadata_name))
|
|
goto_out;
|
|
|
|
if (!(metadata_lv = find_lv(vg, metadata_name))) {
|
|
log_error("Metadata LV %s not found.", metadata_name);
|
|
goto out;
|
|
}
|
|
|
|
if (metadata_lv == lv) {
|
|
log_error("Can't use same LV for pool data and metadata LV %s.",
|
|
display_lvname(metadata_lv));
|
|
goto out;
|
|
}
|
|
|
|
if (!_lvconvert_swap_pool_metadata(cmd, lv, metadata_lv))
|
|
goto_out;
|
|
|
|
return ECMD_PROCESSED;
|
|
|
|
out:
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
int lvconvert_swap_pool_metadata_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_swap_pool_metadata_single);
|
|
}
|
|
|
|
static int _lvconvert_to_pool_or_swap_metadata_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct dm_list *use_pvh = NULL;
|
|
int to_thinpool = 0;
|
|
int to_cachepool = 0;
|
|
int lvt_enum = get_lvt_enum(lv);
|
|
const struct lv_type *lvtype;
|
|
|
|
switch (cmd->command->command_enum) {
|
|
case lvconvert_to_thinpool_or_swap_metadata_CMD:
|
|
if (lv_is_cache(lv) || lv_is_writecache(lv))
|
|
/* For cached LV check the cache origin LV type */
|
|
lvt_enum = get_lvt_enum(seg_lv(first_seg(lv), 0));
|
|
to_thinpool = 1;
|
|
break;
|
|
case lvconvert_to_cachepool_or_swap_metadata_CMD:
|
|
if (lv_is_cache(lv))
|
|
goto_bad; /* Cache over cache is not supported */
|
|
to_cachepool = 1;
|
|
break;
|
|
default:
|
|
log_error(INTERNAL_ERROR "Invalid lvconvert pool command.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
switch (lvt_enum) {
|
|
case thinpool_LVT:
|
|
if (!to_thinpool)
|
|
goto_bad; /* can't accept cache-pool */
|
|
break; /* swap thin-pool */
|
|
case cachepool_LVT:
|
|
if (!to_cachepool)
|
|
goto_bad; /* can't accept thin-pool */
|
|
break; /* swap cache-pool */
|
|
case linear_LVT:
|
|
case raid_LVT:
|
|
case striped_LVT:
|
|
case error_LVT:
|
|
case zero_LVT:
|
|
case vdo_LVT:
|
|
break;
|
|
default:
|
|
bad:
|
|
lvtype = get_lv_type(lvt_enum);
|
|
log_error("LV %s with type %s cannot be used as a %s pool LV.",
|
|
display_lvname(lv), lvtype ? lvtype->name : "unknown",
|
|
to_thinpool ? "thin" : "cache");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
if (lv_is_origin(lv)) {
|
|
log_error("Cannot convert logical volume %s under snapshot.",
|
|
display_lvname(lv));
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
if (!lv_is_visible(lv)) {
|
|
log_error("Can't convert internal LV %s.",
|
|
display_lvname(lv));
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
if (lv_is_locked(lv)) {
|
|
log_error("Can't convert locked LV %s.",
|
|
display_lvname(lv));
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
|
|
return_ECMD_FAILED;
|
|
} else
|
|
use_pvh = &lv->vg->pvs;
|
|
|
|
/*
|
|
* We can finally determine if this command is supposed to create
|
|
* a pool or swap the metadata in an existing pool.
|
|
*
|
|
* This allows the ambiguous command:
|
|
* 'lvconvert --thinpool LV1 --poolmetadata LV2' to mean either:
|
|
* 1. convert LV2 to a pool using the specified meta LV2
|
|
* 2. swap the meta lv in LV1 with LV2
|
|
*
|
|
* In case 2, the poolmetadata option is required, but in case 1
|
|
* it is optional. So, the command def is not able to validate
|
|
* the required/optional option, and we have to check here
|
|
* for missing poolmetadata in case 2.
|
|
*/
|
|
if (lv_is_pool(lv)) {
|
|
if (!arg_is_set(cmd, poolmetadata_ARG)) {
|
|
log_error("The --poolmetadata option is required to swap metadata.");
|
|
return ECMD_FAILED;
|
|
}
|
|
return _lvconvert_swap_pool_metadata_single(cmd, lv, handle);
|
|
}
|
|
|
|
if (!_lvconvert_to_pool(cmd, lv, lv, to_thinpool, to_cachepool, 0, use_pvh))
|
|
return_ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
/*
|
|
* In the command variants with no position LV arg, the LV arg is taken from
|
|
* the --thinpool/--cachepool arg, and the position args are modified to match
|
|
* the standard command form.
|
|
*/
|
|
|
|
int lvconvert_to_pool_or_swap_metadata_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
char *pool_data_name;
|
|
int i, p;
|
|
|
|
switch (cmd->command->command_enum) {
|
|
case lvconvert_to_thinpool_or_swap_metadata_CMD:
|
|
pool_data_name = (char *)arg_str_value(cmd, thinpool_ARG, NULL);
|
|
break;
|
|
case lvconvert_to_cachepool_or_swap_metadata_CMD:
|
|
pool_data_name = (char *)arg_str_value(cmd, cachepool_ARG, NULL);
|
|
break;
|
|
default:
|
|
log_error(INTERNAL_ERROR "Unknown pool conversion.");
|
|
return 0;
|
|
};
|
|
|
|
/* Make the LV the first position arg. */
|
|
|
|
p = cmd->position_argc;
|
|
for (i = 0; i < cmd->position_argc; i++)
|
|
cmd->position_argv[p] = cmd->position_argv[p-1];
|
|
|
|
cmd->position_argv[0] = pool_data_name;
|
|
cmd->position_argc++;
|
|
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_to_pool_or_swap_metadata_single);
|
|
}
|
|
|
|
static int _lvconvert_merge_thin_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
if (!_lvconvert_merge_thin_snapshot(cmd, lv))
|
|
return_ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_merge_thin_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
return process_each_lv(cmd, cmd->position_argc, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_merge_thin_single);
|
|
}
|
|
|
|
static int _lvconvert_detach_writecache(struct cmd_context *cmd, struct processing_handle *handle,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *lv_fast);
|
|
static int _lvconvert_detach_writecache_when_clean(struct cmd_context *cmd,
|
|
struct lvconvert_result *lr);
|
|
|
|
static int _lvconvert_split_cache_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct logical_volume *lv_main = NULL;
|
|
struct logical_volume *lv_fast = NULL;
|
|
struct lv_segment *seg;
|
|
int ret = 0;
|
|
|
|
if (lv_is_writecache(lv)) {
|
|
lv_main = lv;
|
|
lv_fast = first_seg(lv_main)->writecache;
|
|
|
|
} else if (lv_is_cache(lv)) {
|
|
lv_main = lv;
|
|
lv_fast = first_seg(lv_main)->pool_lv;
|
|
|
|
} else if (lv_is_cache_pool(lv)) {
|
|
lv_fast = lv;
|
|
|
|
if ((dm_list_size(&lv_fast->segs_using_this_lv) == 1) &&
|
|
(seg = get_only_segment_using_this_lv(lv_fast)) &&
|
|
seg_is_cache(seg))
|
|
lv_main = seg->lv;
|
|
|
|
} else if (lv_is_thin_pool(lv)) {
|
|
lv_main = seg_lv(first_seg(lv), 0); /* cached _tdata */
|
|
lv_fast = first_seg(lv_main)->pool_lv;
|
|
|
|
} else if (lv_is_vdo_pool(lv)) {
|
|
lv_main = seg_lv(first_seg(lv), 0); /* cached _vdata */
|
|
lv_fast = first_seg(lv_main)->pool_lv;
|
|
}
|
|
|
|
if (!lv_main) {
|
|
log_error("Cannot find LV with cache from %s.", display_lvname(lv));
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
if (!lv_fast) {
|
|
log_error("Cannot find cache %s.", display_lvname(lv));
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
/* If LV is inactive here, ensure it's not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv_main, "ex", 0))
|
|
return_ECMD_FAILED;
|
|
|
|
if (lv_is_writecache(lv_main)) {
|
|
if (!_lvconvert_detach_writecache(cmd, handle, lv_main, lv_fast))
|
|
return_ECMD_FAILED;
|
|
|
|
if (cmd->command->command_enum == lvconvert_split_and_remove_cache_CMD) {
|
|
struct lvconvert_result *lr = (struct lvconvert_result *) handle->custom_handle;
|
|
/*
|
|
* If detach is ongoing, then the remove needs to wait
|
|
* until _lvconvert_detach_writecache_when_clean(),
|
|
* after the detach has finished. When lr->remove_cache
|
|
* has been set, when_clean() knows it should remove
|
|
* lv_fast at the end.
|
|
*/
|
|
if (!lr->wait_cleaner_writecache) {
|
|
if (lvremove_single(cmd, lv_fast, NULL) != ECMD_PROCESSED)
|
|
return_ECMD_FAILED;
|
|
}
|
|
}
|
|
ret = 1;
|
|
} else if (lv_is_cache(lv_main) && lv_is_cache_vol(lv_fast)) {
|
|
if (cmd->command->command_enum == lvconvert_split_and_remove_cache_CMD) {
|
|
ret = _lvconvert_split_and_remove_cachevol(cmd, lv_main, lv_fast);
|
|
|
|
log_print_unless_silent("Logical volume %s is not cached and %s is removed.",
|
|
display_lvname(lv), display_lvname(lv_fast));
|
|
|
|
} else if (cmd->command->command_enum == lvconvert_split_and_keep_cache_CMD) {
|
|
ret = _lvconvert_split_and_keep_cachevol(cmd, lv_main, lv_fast);
|
|
|
|
log_print_unless_silent("Logical volume %s is not cached and %s is unused.",
|
|
display_lvname(lv), display_lvname(lv_fast));
|
|
|
|
} else
|
|
log_error(INTERNAL_ERROR "Unknown cache split command.");
|
|
|
|
} else if (lv_is_cache(lv_main) && lv_is_cache_pool(lv_fast)) {
|
|
if (cmd->command->command_enum == lvconvert_split_and_remove_cache_CMD)
|
|
ret = _lvconvert_split_and_remove_cachepool(cmd, lv_main, lv_fast);
|
|
|
|
else if (cmd->command->command_enum == lvconvert_split_and_keep_cache_CMD)
|
|
ret = _lvconvert_split_and_keep_cachepool(cmd, lv_main, lv_fast);
|
|
|
|
else
|
|
log_error(INTERNAL_ERROR "Unknown cache split command.");
|
|
} else
|
|
log_error(INTERNAL_ERROR "Unknown cache split command.");
|
|
|
|
if (!ret)
|
|
return ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_split_cache_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_result lr = { 0 };
|
|
int ret;
|
|
|
|
cmd->handles_missing_pvs = 1;
|
|
cmd->partial_activation = 1;
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lr;
|
|
|
|
ret = process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
handle, NULL, &_lvconvert_split_cache_single);
|
|
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
if (ret == ECMD_FAILED)
|
|
return ret;
|
|
|
|
if (lr.wait_cleaner_writecache)
|
|
if (!_lvconvert_detach_writecache_when_clean(cmd, &lr))
|
|
ret = ECMD_FAILED;
|
|
|
|
return ret;
|
|
}
|
|
|
|
static int _lvconvert_raid_types_single(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct lvconvert_params *lp = (struct lvconvert_params *) handle->custom_handle;
|
|
struct dm_list *use_pvh;
|
|
struct convert_poll_id_list *idl;
|
|
int ret;
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
|
|
return_ECMD_FAILED;
|
|
lp->pv_count = cmd->position_argc - 1;
|
|
} else
|
|
use_pvh = &lv->vg->pvs;
|
|
|
|
lp->pvh = use_pvh;
|
|
|
|
lp->lv_to_poll = lv;
|
|
|
|
ret = _lvconvert_raid_types(cmd, lv, lp);
|
|
|
|
if (ret != ECMD_PROCESSED)
|
|
return_ECMD_FAILED;
|
|
|
|
if (lp->need_polling) {
|
|
/* _lvconvert() call may alter the reference in lp->lv_to_poll */
|
|
if (!lv_is_active(lp->lv_to_poll))
|
|
log_print_unless_silent("Conversion starts after activation.");
|
|
else {
|
|
if (!(idl = _convert_poll_id_list_create(cmd, lp->lv_to_poll)))
|
|
return_ECMD_FAILED;
|
|
dm_list_add(&lp->idls, &idl->list);
|
|
}
|
|
}
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
static int _lvconvert_raid_types_check(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle,
|
|
int lv_is_named_arg)
|
|
{
|
|
int lvt_enum = get_lvt_enum(lv);
|
|
const struct lv_type *lvtype = get_lv_type(lvt_enum);
|
|
|
|
if (!lv_is_visible(lv)) {
|
|
if (!lv_is_cache_pool_metadata(lv) &&
|
|
!lv_is_cache_pool_data(lv) &&
|
|
!lv_is_thin_pool_metadata(lv) &&
|
|
!lv_is_thin_pool_data(lv) &&
|
|
!lv_is_vdo_pool_data(lv) &&
|
|
!lv_is_used_cache_pool(lv) &&
|
|
!lv_is_mirrored(lv) &&
|
|
!lv_is_raid(lv))
|
|
goto fail_hidden;
|
|
}
|
|
|
|
/*
|
|
* FIXME: this validation could be done by command defs.
|
|
*
|
|
* Outside the standard linear/striped/mirror/raid LV
|
|
* types, cache is the only special LV type that is handled
|
|
* (the command is redirected to origin).
|
|
*/
|
|
switch (lvt_enum) {
|
|
case thin_LVT:
|
|
case thinpool_LVT:
|
|
case cachepool_LVT:
|
|
case snapshot_LVT:
|
|
log_error("Operation not permitted on LV %s type %s.",
|
|
display_lvname(lv), lvtype ? lvtype->name : "unknown");
|
|
return 0;
|
|
}
|
|
|
|
return 1;
|
|
|
|
fail_hidden:
|
|
log_error("Operation not permitted on hidden LV %s.", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
int lvconvert_raid_types_cmd(struct cmd_context * cmd, int argc, char **argv)
|
|
{
|
|
int poll_ret, ret;
|
|
int saved_ignore_suspended_devices;
|
|
struct processing_handle *handle;
|
|
struct convert_poll_id_list *idl;
|
|
struct lvconvert_params lp = {
|
|
.conv_type = CONV_OTHER,
|
|
.target_attr = ~0,
|
|
.idls = DM_LIST_HEAD_INIT(lp.idls),
|
|
};
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lp;
|
|
|
|
if (!_read_params(cmd, &lp)) {
|
|
ret = EINVALID_CMD_LINE;
|
|
goto_out;
|
|
}
|
|
|
|
saved_ignore_suspended_devices = ignore_suspended_devices();
|
|
|
|
ret = process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
handle, &_lvconvert_raid_types_check, &_lvconvert_raid_types_single);
|
|
|
|
init_ignore_suspended_devices(saved_ignore_suspended_devices);
|
|
|
|
dm_list_iterate_items(idl, &lp.idls) {
|
|
poll_ret = _lvconvert_poll_by_id(cmd, idl->id,
|
|
lp.wait_completion ? 0 : 1U,
|
|
idl->is_merging_origin,
|
|
idl->is_merging_origin_thin);
|
|
if (poll_ret > ret)
|
|
ret = poll_ret;
|
|
}
|
|
|
|
out:
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
/*
|
|
* change mirror log
|
|
*/
|
|
|
|
static int _lvconvert_visible_check(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle,
|
|
int lv_is_named_arg)
|
|
{
|
|
if (!lv_is_visible(lv)) {
|
|
log_error("Operation not permitted on hidden LV %s.", display_lvname(lv));
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
static int _lvconvert_change_mirrorlog_single(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct lvconvert_params *lp = (struct lvconvert_params *) handle->custom_handle;
|
|
struct dm_list *use_pvh;
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
|
|
return_ECMD_FAILED;
|
|
lp->pv_count = cmd->position_argc - 1;
|
|
} else
|
|
use_pvh = &lv->vg->pvs;
|
|
|
|
lp->pvh = use_pvh;
|
|
|
|
/* FIXME: extract the mirrorlog functionality out of _lvconvert_raid_types()? */
|
|
return _lvconvert_raid_types(cmd, lv, lp);
|
|
}
|
|
|
|
int lvconvert_change_mirrorlog_cmd(struct cmd_context * cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_params lp = {
|
|
.conv_type = CONV_OTHER,
|
|
.target_attr = ~0,
|
|
.idls = DM_LIST_HEAD_INIT(lp.idls),
|
|
};
|
|
int ret;
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lp;
|
|
|
|
/* FIXME: extract the relevant bits of read_params and put here. */
|
|
if (!_read_params(cmd, &lp)) {
|
|
ret = EINVALID_CMD_LINE;
|
|
goto_out;
|
|
}
|
|
|
|
ret = process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
handle, &_lvconvert_visible_check, &_lvconvert_change_mirrorlog_single);
|
|
|
|
out:
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
static int _lvconvert_change_region_size_single(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
if (!lv_raid_change_region_size(lv, arg_is_set(cmd, yes_ARG), arg_count(cmd, force_ARG),
|
|
arg_int_value(cmd, regionsize_ARG, 0)))
|
|
return_ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_change_region_size_cmd(struct cmd_context * cmd, int argc, char **argv)
|
|
{
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, &_lvconvert_visible_check, &_lvconvert_change_region_size_single);
|
|
}
|
|
|
|
/*
|
|
* split mirror images
|
|
*/
|
|
|
|
static int _lvconvert_split_mirror_images_single(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct lvconvert_params *lp = (struct lvconvert_params *) handle->custom_handle;
|
|
struct dm_list *use_pvh;
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
|
|
return_ECMD_FAILED;
|
|
lp->pv_count = cmd->position_argc - 1;
|
|
} else
|
|
use_pvh = &lv->vg->pvs;
|
|
|
|
lp->pvh = use_pvh;
|
|
|
|
/* FIXME: extract the split functionality out of _lvconvert_raid_types()? */
|
|
return _lvconvert_raid_types(cmd, lv, lp);
|
|
}
|
|
|
|
int lvconvert_split_mirror_images_cmd(struct cmd_context * cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_params lp = {
|
|
.conv_type = CONV_OTHER,
|
|
.target_attr = ~0,
|
|
.idls = DM_LIST_HEAD_INIT(lp.idls),
|
|
};
|
|
int ret;
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lp;
|
|
|
|
/* FIXME: extract the relevant bits of read_params and put here. */
|
|
if (!_read_params(cmd, &lp)) {
|
|
ret = EINVALID_CMD_LINE;
|
|
goto_out;
|
|
}
|
|
|
|
/* FIXME: are there any hidden LVs that should be disallowed? */
|
|
|
|
ret = process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
handle, NULL, &_lvconvert_split_mirror_images_single);
|
|
|
|
out:
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
/*
|
|
* merge mirror images
|
|
*
|
|
* Called from both lvconvert --mergemirrors and lvconvert --merge.
|
|
*/
|
|
|
|
static int _lvconvert_merge_mirror_images_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
if (!lv_raid_merge(lv))
|
|
return_ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_merge_mirror_images_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
/* arg can be a VG name, which is the standard option usage */
|
|
cmd->get_vgname_from_options = 0;
|
|
|
|
return process_each_lv(cmd, cmd->position_argc, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, &_lvconvert_visible_check, &_lvconvert_merge_mirror_images_single);
|
|
}
|
|
|
|
static int _lvconvert_merge_generic_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
int ret;
|
|
|
|
if (lv_is_cow(lv))
|
|
ret = _lvconvert_merge_snapshot_single(cmd, lv, handle);
|
|
|
|
else if (lv_is_thin_volume(lv))
|
|
ret = _lvconvert_merge_thin_single(cmd, lv, handle);
|
|
|
|
else
|
|
ret = _lvconvert_merge_mirror_images_single(cmd, lv, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
int lvconvert_merge_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_result lr = { 0 };
|
|
struct convert_poll_id_list *idl;
|
|
int ret, poll_ret;
|
|
|
|
dm_list_init(&lr.poll_idls);
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lr;
|
|
|
|
cmd->get_vgname_from_options = 0;
|
|
|
|
ret = process_each_lv(cmd, cmd->position_argc, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
handle, NULL, &_lvconvert_merge_generic_single);
|
|
|
|
/* polling is only used by merge_snapshot */
|
|
if (lr.need_polling) {
|
|
dm_list_iterate_items(idl, &lr.poll_idls) {
|
|
poll_ret = _lvconvert_poll_by_id(cmd, idl->id,
|
|
arg_is_set(cmd, background_ARG), 1, 0);
|
|
if (poll_ret > ret)
|
|
ret = poll_ret;
|
|
}
|
|
}
|
|
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
static int _lvconvert_to_vdopool_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *vdo_lv;
|
|
const char *vg_name = NULL;
|
|
struct vdo_convert_params vcp = {
|
|
.activate = CHANGE_AEY,
|
|
.lv_name = arg_str_value(cmd, name_ARG, NULL),
|
|
.virtual_extents = extents_from_size(cmd,
|
|
arg_uint64_value(cmd, virtualsize_ARG, UINT64_C(0)),
|
|
vg->extent_size),
|
|
.do_zero = arg_int_value(cmd, zero_ARG, 1),
|
|
.do_wipe_signatures = 1,
|
|
.yes = arg_count(cmd, yes_ARG),
|
|
.force = arg_count(cmd, force_ARG)
|
|
};
|
|
|
|
if (vcp.lv_name) {
|
|
if (!validate_restricted_lvname_param(cmd, &vg_name, &vcp.lv_name))
|
|
goto_out;
|
|
} else
|
|
vcp.lv_name = "lvol%d";
|
|
|
|
if (!fill_vdo_target_params(cmd, &vcp.vdo_params, &vcp.header_size, vg->profile))
|
|
goto_out;
|
|
|
|
if (!get_vdo_settings(cmd, &vcp.vdo_params, NULL))
|
|
goto_out;
|
|
|
|
/* If LV is inactive here, ensure it's not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
goto_out;
|
|
|
|
log_warn("WARNING: Converting logical volume %s to VDO pool volume %s formatting.",
|
|
display_lvname(lv), vcp.do_zero ? "with" : "WITHOUT");
|
|
|
|
if (vcp.do_zero)
|
|
log_warn("THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)");
|
|
else
|
|
log_warn("WARNING: Using invalid VDO pool data MAY DESTROY YOUR DATA!");
|
|
|
|
if (!vcp.yes &&
|
|
yes_no_prompt("Do you really want to convert %s? [y/n]: ",
|
|
display_lvname(lv)) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
goto out;
|
|
}
|
|
|
|
if (!(vdo_lv = convert_vdo_lv(lv, &vcp)))
|
|
goto_out;
|
|
|
|
log_print_unless_silent("Converted %s to VDO pool volume and created virtual %s VDO volume.",
|
|
display_lvname(lv), display_lvname(vdo_lv));
|
|
|
|
return ECMD_PROCESSED;
|
|
|
|
out:
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
int lvconvert_to_vdopool_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_to_vdopool_single);
|
|
}
|
|
|
|
int lvconvert_to_vdopool_param_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
/* Make the LV the first position arg. */
|
|
int i, p = cmd->position_argc;
|
|
|
|
for (i = 0; i < cmd->position_argc; i++)
|
|
cmd->position_argv[p] = cmd->position_argv[p-1];
|
|
|
|
cmd->position_argv[0] = (char *)arg_str_value(cmd, vdopool_ARG, NULL);
|
|
cmd->position_argc++;
|
|
|
|
return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
|
|
NULL, NULL, &_lvconvert_to_vdopool_single);
|
|
}
|
|
|
|
/*
|
|
* Starts the detach process, and may complete it, or may defer the completion
|
|
* if cleaning is required, by returning a poll id. If deferred, the caller
|
|
* will notice the poll id and call lvconvert_detach_writecache_when_clean
|
|
* to wait for the cleaning and complete the detach. The command can be cancelled
|
|
* while waiting for cleaning and the same command be repeated to continue the
|
|
* process.
|
|
*/
|
|
static int _lvconvert_detach_writecache(struct cmd_context *cmd,
|
|
struct processing_handle *handle,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *lv_fast)
|
|
{
|
|
struct lvconvert_result *lr = (struct lvconvert_result *) handle->custom_handle;
|
|
struct writecache_settings settings;
|
|
struct convert_poll_id_list *idl;
|
|
uint32_t block_size_sectors;
|
|
int active_begin = 0;
|
|
int active_clean = 0;
|
|
int is_clean = 0;
|
|
int noflush = 0;
|
|
|
|
dm_list_init(&lr->poll_idls);
|
|
|
|
memset(&settings, 0, sizeof(settings));
|
|
|
|
if (!get_writecache_settings(cmd, &settings, &block_size_sectors)) {
|
|
log_error("Invalid writecache settings.");
|
|
return 0;
|
|
}
|
|
|
|
if (!archive(lv->vg))
|
|
return_0;
|
|
|
|
/*
|
|
* If the LV is inactive when we begin, then we want to
|
|
* deactivate the LV at the end.
|
|
*/
|
|
active_begin = lv_is_active(lv);
|
|
|
|
if (lv_is_partial(lv_fast) || (!active_begin && arg_count(cmd, force_ARG))) {
|
|
if (!arg_count(cmd, force_ARG)) {
|
|
log_warn("WARNING: writecache on %s is not complete and cannot be flushed.", display_lvname(lv_fast));
|
|
log_warn("WARNING: cannot detach writecache from %s without --force.", display_lvname(lv));
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
|
|
log_warn("WARNING: Data may be lost by detaching writecache without flushing.");
|
|
|
|
if (!arg_count(cmd, yes_ARG) &&
|
|
yes_no_prompt("Detach writecache %s from %s without flushing data?",
|
|
display_lvname(lv_fast), display_lvname(lv)) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
|
|
noflush = 1;
|
|
}
|
|
|
|
if (!noflush) {
|
|
/*
|
|
* --cachesettings cleaner=0 means to skip the use of the cleaner
|
|
* and go directly to detach which will use a flush message.
|
|
* (This is currently the only cachesetting used during detach.)
|
|
*/
|
|
if (settings.cleaner_set && !settings.cleaner) {
|
|
log_print_unless_silent("Detaching writecache skipping cleaner...");
|
|
goto detach;
|
|
}
|
|
|
|
if (!writecache_cleaner_supported(cmd)) {
|
|
log_print_unless_silent("Detaching writecache without cleaner...");
|
|
goto detach;
|
|
}
|
|
|
|
if (!active_begin && !activate_lv(cmd, lv)) {
|
|
log_error("Failed to activate LV to clean writecache.");
|
|
return 0;
|
|
}
|
|
active_clean = 1;
|
|
|
|
/*
|
|
* If the user ran this command previously (or set cleaner
|
|
* directly) the cache may already be empty and ready for
|
|
* detach.
|
|
*/
|
|
if (lv_writecache_is_clean(cmd, lv, NULL)) {
|
|
log_print_unless_silent("Detaching writecache already clean.");
|
|
is_clean = 1;
|
|
goto detach;
|
|
}
|
|
|
|
/*
|
|
* If the user has not already done lvchange --cachesettings cleaner=1
|
|
* then do that here. If the LV is inactive, this activates it
|
|
* so that cache writeback can be done.
|
|
*/
|
|
log_print_unless_silent("Detaching writecache setting cleaner.");
|
|
|
|
if (!lv_writecache_set_cleaner(lv)) {
|
|
log_error("Failed to set cleaner cachesetting to flush cache.");
|
|
log_error("See lvchange --cachesettings cleaner=1");
|
|
|
|
if (!active_begin && active_clean && !deactivate_lv(cmd, lv))
|
|
stack;
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* The cache may have been nearly clean and will be empty with
|
|
* a short dely.
|
|
*/
|
|
usleep(10000);
|
|
if (lv_writecache_is_clean(cmd, lv, NULL)) {
|
|
log_print_unless_silent("Detaching writecache finished cleaning.");
|
|
is_clean = 1;
|
|
goto detach;
|
|
}
|
|
|
|
if (!(idl = _convert_poll_id_list_create(cmd, lv))) {
|
|
log_error("Failed to monitor writecache cleaner progress.");
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* Monitor the writecache status until the cache is unused.
|
|
* This is done at the end of the command where locks are not
|
|
* held since the writeback can take some time.
|
|
*/
|
|
lr->wait_cleaner_writecache = 1;
|
|
lr->active_begin = active_begin;
|
|
|
|
/* The command wants to remove the cache after detaching. */
|
|
if (cmd->command->command_enum == lvconvert_split_and_remove_cache_CMD)
|
|
lr->remove_cache = 1;
|
|
|
|
dm_list_add(&lr->poll_idls, &idl->list);
|
|
return 1;
|
|
}
|
|
|
|
detach:
|
|
|
|
/*
|
|
* If the LV was inactive before cleaning and activated to do cleaning,
|
|
* then deactivate before the detach.
|
|
*/
|
|
if (!active_begin && active_clean && !deactivate_lv(cmd, lv))
|
|
stack;
|
|
|
|
if (is_clean)
|
|
noflush = 1;
|
|
|
|
if (!lv_detach_writecache_cachevol(lv, noflush))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Logical volume %s writecache has been detached.",
|
|
display_lvname(lv));
|
|
return 1;
|
|
}
|
|
|
|
/*
|
|
* _lvconvert_detach_writecache() set the cleaner option for the LV
|
|
* so writecache will begin writing back data from cache to origin.
|
|
* It then saved the LV name/id (lvconvert_result/poll_id), and
|
|
* exited process_each_lv (releasing the VG and VG lock). Then
|
|
* this is called to monitor the progress of the cache writeback.
|
|
* When the cache is clean, this does the detach (writecache is removed
|
|
* in metadata and LV in kernel is updated.)
|
|
*/
|
|
static int _lvconvert_detach_writecache_when_clean(struct cmd_context *cmd,
|
|
struct lvconvert_result *lr)
|
|
{
|
|
struct convert_poll_id_list *idl;
|
|
struct poll_operation_id *id;
|
|
struct volume_group *vg;
|
|
struct logical_volume *lv;
|
|
struct logical_volume *lv_fast;
|
|
uint32_t lockd_state, error_flags;
|
|
uint64_t dirty;
|
|
int ret = 0;
|
|
|
|
idl = dm_list_item(dm_list_first(&lr->poll_idls), struct convert_poll_id_list);
|
|
id = idl->id;
|
|
|
|
/*
|
|
* TODO: we should be able to save info about the dm device for this LV
|
|
* and monitor the dm device status without doing vg lock/read around
|
|
* each check. The vg lock/read/write would then happen only once when
|
|
* status was finished and we want to finish the detach. If the dm
|
|
* device goes away while monitoring, it's no different and no worse
|
|
* than the LV going away here.
|
|
*/
|
|
|
|
retry:
|
|
lockd_state = 0;
|
|
error_flags = 0;
|
|
|
|
if (!lockd_vg(cmd, id->vg_name, "ex", 0, &lockd_state)) {
|
|
log_error("Detaching writecache interrupted - locking VG failed.");
|
|
return 0;
|
|
}
|
|
|
|
log_debug("detach writecache check clean reading vg %s", id->vg_name);
|
|
|
|
vg = vg_read(cmd, id->vg_name, NULL, READ_FOR_UPDATE, lockd_state, &error_flags, NULL);
|
|
|
|
if (!vg) {
|
|
log_error("Detaching writecache interrupted - reading VG failed.");
|
|
goto out_lockd;
|
|
}
|
|
|
|
if (error_flags) {
|
|
log_error("Detaching writecache interrupted - reading VG error %x.", error_flags);
|
|
goto out_release;
|
|
}
|
|
|
|
lv = find_lv(vg, id->lv_name);
|
|
|
|
if (lv && id->uuid && strcmp(id->uuid, (char *)&lv->lvid))
|
|
lv = NULL;
|
|
|
|
if (!lv) {
|
|
log_error("Detaching writecache interrupted - LV not found.");
|
|
goto out_release;
|
|
}
|
|
|
|
if (!lv_is_active(lv)) {
|
|
log_error("Detaching writecache interrupted - LV not active.");
|
|
goto out_release;
|
|
}
|
|
|
|
if (!lv_writecache_is_clean(cmd, lv, &dirty)) {
|
|
unlock_and_release_vg(cmd, vg, vg->name);
|
|
|
|
if (!lockd_vg(cmd, id->vg_name, "un", 0, &lockd_state))
|
|
stack;
|
|
|
|
log_print_unless_silent("Detaching writecache cleaning %llu blocks", (unsigned long long)dirty);
|
|
log_print_unless_silent("This command can be cancelled and rerun to complete writecache detach.");
|
|
sleep(5);
|
|
goto retry;
|
|
}
|
|
|
|
if (!lr->active_begin) {
|
|
/*
|
|
* The LV was not active to begin so we should leave it inactive at the end.
|
|
* It will remain inactive during detach since it's clean and doesn't need
|
|
* a flush message.
|
|
*/
|
|
if (!deactivate_lv(cmd, lv))
|
|
stack;
|
|
}
|
|
|
|
log_print_unless_silent("Detaching writecache completed cleaning.");
|
|
|
|
lv_fast = first_seg(lv)->writecache;
|
|
|
|
/*
|
|
* When the cleaner has finished, we can detach with noflush since
|
|
* the cleaner has done the flushing.
|
|
*/
|
|
|
|
if (!lv_detach_writecache_cachevol(lv, 1)) {
|
|
log_error("Detaching writecache cachevol failed.");
|
|
goto out_release;
|
|
}
|
|
|
|
/*
|
|
* The detach was started by an uncache command that wants to remove
|
|
* the cachevol after detaching.
|
|
*/
|
|
if (lr->remove_cache) {
|
|
if (lvremove_single(cmd, lv_fast, NULL) != ECMD_PROCESSED) {
|
|
log_error("Removing the writecache cachevol failed.");
|
|
goto out_release;
|
|
}
|
|
}
|
|
|
|
ret = 1;
|
|
|
|
out_release:
|
|
if (ret)
|
|
log_print_unless_silent("Logical volume %s write cache has been detached.", display_lvname(lv));
|
|
|
|
unlock_and_release_vg(cmd, vg, vg->name);
|
|
|
|
out_lockd:
|
|
if (!lockd_vg(cmd, id->vg_name, "un", 0, &lockd_state))
|
|
stack;
|
|
|
|
return ret;
|
|
}
|
|
|
|
static int _writecache_zero(struct cmd_context *cmd, struct logical_volume *lv)
|
|
{
|
|
struct wipe_params wp = {
|
|
.do_wipe_signatures = 1, /* optional, to print warning if clobbering something */
|
|
.do_zero = 1, /* required for dm-writecache to work */
|
|
.yes = arg_count(cmd, yes_ARG),
|
|
.force = arg_count(cmd, force_ARG)
|
|
};
|
|
int ret;
|
|
|
|
if (!(lv->status & LVM_WRITE)) {
|
|
log_error("Cannot initialize readonly LV %s", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (test_mode())
|
|
return 1;
|
|
|
|
if (!activate_lv(cmd, lv)) {
|
|
log_error("Failed to activate LV %s for zeroing.", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (!(ret = wipe_lv(lv, wp)))
|
|
stack;
|
|
|
|
if (!deactivate_lv(cmd, lv)) {
|
|
log_error("Failed to deactivate LV %s for zeroing.", display_lvname(lv));
|
|
ret = 0;
|
|
}
|
|
|
|
return ret;
|
|
}
|
|
|
|
static struct logical_volume *_lv_writecache_create(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct logical_volume *lv_fast,
|
|
uint32_t block_size_sectors,
|
|
struct writecache_settings *settings)
|
|
{
|
|
struct logical_volume *lv_wcorig;
|
|
const struct segment_type *segtype;
|
|
struct lv_segment *seg;
|
|
|
|
/* should lv_fast get a new status flag indicating it's the cache in a writecache LV? */
|
|
|
|
if (!(segtype = get_segtype_from_string(cmd, SEG_TYPE_NAME_WRITECACHE)))
|
|
return_NULL;
|
|
|
|
/*
|
|
* "lv_wcorig" is a new LV with new id, but with the segments from "lv".
|
|
* "lv" keeps the existing name and id, but gets a new writecache segment,
|
|
* in place of the segments that were moved to lv_wcorig.
|
|
*/
|
|
|
|
if (!(lv_wcorig = insert_layer_for_lv(cmd, lv, 0, "_wcorig")))
|
|
return_NULL;
|
|
|
|
lv->status |= WRITECACHE;
|
|
seg = first_seg(lv);
|
|
seg->segtype = segtype;
|
|
|
|
seg->writecache = lv_fast;
|
|
lv_set_hidden(lv_fast);
|
|
|
|
/* writecache_block_size is in bytes */
|
|
seg->writecache_block_size = block_size_sectors * 512;
|
|
|
|
memcpy(&seg->writecache_settings, settings, sizeof(struct writecache_settings));
|
|
|
|
if (!add_seg_to_segs_using_this_lv(lv_fast, seg))
|
|
return_NULL;
|
|
|
|
return lv_wcorig;
|
|
}
|
|
|
|
/*
|
|
* Currently only supports writecache block sizes 512 and 4096.
|
|
* This could be expanded later.
|
|
*/
|
|
static int _set_writecache_block_size(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
uint32_t *block_size_sectors)
|
|
{
|
|
char pathname[PATH_MAX];
|
|
struct dm_list pvs_list;
|
|
struct pv_list *pvl;
|
|
uint32_t fs_block_size = 0;
|
|
uint32_t block_size_setting = 0;
|
|
uint32_t block_size = 0;
|
|
int lbs_unknown = 0, lbs_4k = 0, lbs_512 = 0;
|
|
int pbs_unknown = 0, pbs_4k = 0, pbs_512 = 0;
|
|
int rv = 0;
|
|
|
|
/* This is set if the user specified a writecache block size on the command line. */
|
|
if (*block_size_sectors)
|
|
block_size_setting = *block_size_sectors * 512;
|
|
|
|
dm_list_init(&pvs_list);
|
|
|
|
if (!get_pv_list_for_lv(cmd->mem, lv, &pvs_list)) {
|
|
log_error("Failed to build list of PVs for %s.", display_lvname(lv));
|
|
goto bad;
|
|
}
|
|
|
|
dm_list_iterate_items(pvl, &pvs_list) {
|
|
unsigned int pbs = 0;
|
|
unsigned int lbs = 0;
|
|
|
|
if (!dev_get_direct_block_sizes(pvl->pv->dev, &pbs, &lbs)) {
|
|
lbs_unknown++;
|
|
pbs_unknown++;
|
|
continue;
|
|
}
|
|
|
|
if (lbs == 4096)
|
|
lbs_4k++;
|
|
else if (lbs == 512)
|
|
lbs_512++;
|
|
else
|
|
lbs_unknown++;
|
|
|
|
if (pbs == 4096)
|
|
pbs_4k++;
|
|
else if (pbs == 512)
|
|
pbs_512++;
|
|
else
|
|
pbs_unknown++;
|
|
}
|
|
|
|
if (lbs_4k && lbs_512) {
|
|
log_error("Writecache requires consistent logical block size for LV devices.");
|
|
goto bad;
|
|
}
|
|
|
|
if (lbs_4k && block_size_setting && (block_size_setting < 4096)) {
|
|
log_error("Writecache block size %u not allowed with device logical block size 4096.",
|
|
block_size_setting);
|
|
goto bad;
|
|
}
|
|
|
|
/*
|
|
* When attaching writecache to thin pool data, the fs block sizes
|
|
* would need to be checked on each thin LV which isn't practical, so
|
|
* default to 512, and require the user to specify 4k when appropriate.
|
|
*/
|
|
if (lv_is_thin_pool(lv) || lv_is_thin_pool_data(lv)) {
|
|
if (block_size_setting)
|
|
block_size = block_size_setting;
|
|
else
|
|
block_size = 512;
|
|
|
|
log_print_unless_silent("Using writecache block size %u for thin pool data, logical block size %u, physical block size %u.",
|
|
block_size, lbs_4k ? 4096 : 512, pbs_4k ? 4096 : 512);
|
|
|
|
goto out;
|
|
}
|
|
|
|
if (dm_snprintf(pathname, sizeof(pathname), "%s%s/%s",
|
|
cmd->dev_dir, lv->vg->name, lv->name) < 0) {
|
|
log_error("Path name too long to get LV block size %s", display_lvname(lv));
|
|
goto bad;
|
|
}
|
|
|
|
if (test_mode()) {
|
|
log_print_unless_silent("Test mode skips checking fs block size.");
|
|
fs_block_size = 0;
|
|
goto skip_fs;
|
|
}
|
|
|
|
/*
|
|
* fs_block_size_and_type() returns the libblkid BLOCK_SIZE value,
|
|
* where libblkid has fs-specific code to set BLOCK_SIZE to the
|
|
* value we need here.
|
|
*
|
|
* The term "block size" here may not equate directly to what the fs
|
|
* calls the block size, e.g. xfs calls this the sector size (and
|
|
* something different the block size); while ext4 does call this
|
|
* value the block size, but it's possible values are not the same
|
|
* as xfs's, and do not seem to relate directly to the device LBS.
|
|
*
|
|
* With 512 LBS and 4K PBS, mkfs.xfs will use xfs sector size 4K.
|
|
*/
|
|
rv = fs_block_size_and_type(pathname, &fs_block_size, NULL, NULL);
|
|
skip_fs:
|
|
if (!rv || !fs_block_size) {
|
|
if (block_size_setting)
|
|
block_size = block_size_setting;
|
|
else
|
|
block_size = 4096;
|
|
|
|
log_print_unless_silent("Using writecache block size %u for unknown file system block size, logical block size %u, physical block size %u.",
|
|
block_size, lbs_4k ? 4096 : 512, pbs_4k ? 4096 : 512);
|
|
|
|
if (block_size != 512) {
|
|
log_warn("WARNING: unable to detect a file system block size on %s", display_lvname(lv));
|
|
log_warn("WARNING: using a writecache block size larger than the file system block size may corrupt the file system.");
|
|
if (!arg_is_set(cmd, yes_ARG) &&
|
|
yes_no_prompt("Use writecache block size %u? [y/n]: ", block_size) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
goto bad;
|
|
}
|
|
}
|
|
|
|
goto out;
|
|
}
|
|
|
|
if (!block_size_setting) {
|
|
/* User did not specify a block size, so choose according to fs block size. */
|
|
if (fs_block_size == 4096)
|
|
block_size = 4096;
|
|
else if (fs_block_size == 512)
|
|
block_size = 512;
|
|
else if (fs_block_size > 4096)
|
|
block_size = 4096;
|
|
else if (fs_block_size < 4096)
|
|
block_size = 512;
|
|
else
|
|
goto_bad;
|
|
} else {
|
|
if (block_size_setting <= fs_block_size)
|
|
block_size = block_size_setting;
|
|
else {
|
|
log_error("Writecache block size %u cannot be larger than file system block size %u.",
|
|
block_size_setting, fs_block_size);
|
|
goto bad;
|
|
}
|
|
}
|
|
|
|
out:
|
|
if (block_size == 512)
|
|
*block_size_sectors = 1;
|
|
else if (block_size == 4096)
|
|
*block_size_sectors = 8;
|
|
else
|
|
goto_bad;
|
|
|
|
return 1;
|
|
bad:
|
|
return 0;
|
|
}
|
|
|
|
static int _check_writecache_memory(struct cmd_context *cmd, struct logical_volume *lv_fast,
|
|
uint32_t block_size_sectors)
|
|
{
|
|
char line[128];
|
|
FILE *fp;
|
|
uint64_t cachevol_size_bytes = lv_fast->size * SECTOR_SIZE;
|
|
uint64_t need_mem_bytes = 0;
|
|
uint64_t proc_mem_bytes = 0;
|
|
uint64_t need_mem_gb;
|
|
uint64_t proc_mem_gb;
|
|
unsigned long long proc_mem_kb = 0;
|
|
char proc_meminfo[PATH_MAX];
|
|
|
|
if (*cmd->proc_dir)
|
|
goto skip_proc;
|
|
|
|
if (dm_snprintf(proc_meminfo, sizeof(proc_meminfo),
|
|
"%s/meminfo", cmd->proc_dir) < 0) {
|
|
stack;
|
|
goto skip_proc;
|
|
}
|
|
|
|
if (!(fp = fopen(proc_meminfo, "r")))
|
|
goto skip_proc;
|
|
|
|
while (fgets(line, sizeof(line), fp)) {
|
|
if (strncmp(line, "MemTotal:", 9))
|
|
continue;
|
|
if (sscanf(line, "%*s%llu%*s", &proc_mem_kb) != 1)
|
|
break;
|
|
break;
|
|
}
|
|
(void)fclose(fp);
|
|
|
|
proc_mem_bytes = proc_mem_kb * 1024;
|
|
|
|
skip_proc:
|
|
/* dm-writecache memory consumption per block is 88 bytes */
|
|
if (block_size_sectors == 8) {
|
|
need_mem_bytes = cachevol_size_bytes * 88 / 4096;
|
|
} else if (block_size_sectors == 1) {
|
|
need_mem_bytes = cachevol_size_bytes * 88 / 512;
|
|
} else {
|
|
/* shouldn't happen */
|
|
log_warn("Unknown memory usage for unknown writecache block_size_sectors %u", block_size_sectors);
|
|
return 1;
|
|
}
|
|
|
|
need_mem_gb = need_mem_bytes / 1073741824;
|
|
proc_mem_gb = proc_mem_bytes / 1073741824;
|
|
|
|
/*
|
|
* warn if writecache needs > 50% of main memory, and
|
|
* confirm if writecache needs > 90% of main memory.
|
|
*/
|
|
if (need_mem_bytes >= (proc_mem_bytes / 2)) {
|
|
log_warn("WARNING: writecache size %s will use %llu GiB of system memory (%llu GiB).",
|
|
display_size(cmd, lv_fast->size),
|
|
(unsigned long long)need_mem_gb,
|
|
(unsigned long long)proc_mem_gb);
|
|
|
|
if (need_mem_gb >= (proc_mem_gb * 9 / 10)) {
|
|
if (!arg_is_set(cmd, yes_ARG) &&
|
|
yes_no_prompt("Continue adding writecache? [y/n]: ") == 'n') {
|
|
log_error("Conversion aborted.");
|
|
return 0;
|
|
}
|
|
}
|
|
}
|
|
|
|
return 1;
|
|
}
|
|
|
|
int lvconvert_writecache_attach_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct logical_volume *lv_update;
|
|
struct logical_volume *lv_wcorig;
|
|
struct logical_volume *lv_fast;
|
|
struct writecache_settings settings = { 0 };
|
|
const char *fast_name;
|
|
uint32_t block_size_sectors = 0;
|
|
char *lockd_fast_args = NULL;
|
|
char *lockd_fast_name = NULL;
|
|
struct id lockd_fast_id;
|
|
char cvol_name[NAME_LEN];
|
|
int is_active;
|
|
|
|
/*
|
|
* User specifies an existing cachevol to use or a cachedevice
|
|
* to create a cachevol from.
|
|
*/
|
|
if ((fast_name = arg_str_value(cmd, cachevol_ARG, NULL))) {
|
|
if (!validate_lvname_param(cmd, &vg->name, &fast_name))
|
|
goto_bad;
|
|
|
|
if (!(lv_fast = find_lv(vg, fast_name))) {
|
|
log_error("LV %s not found.", fast_name);
|
|
goto bad;
|
|
}
|
|
|
|
if (lv_fast == lv) {
|
|
log_error("Invalid cachevol LV.");
|
|
goto bad;
|
|
}
|
|
|
|
if (lv_is_cache_vol(lv_fast)) {
|
|
log_error("LV %s is already used as a cachevol.", display_lvname(lv_fast));
|
|
goto bad;
|
|
}
|
|
|
|
if (!seg_is_linear(first_seg(lv_fast))) {
|
|
log_error("LV %s must be linear to use as a writecache.", display_lvname(lv_fast));
|
|
goto bad;
|
|
}
|
|
|
|
/* fast LV shouldn't generally be active by itself, but just in case. */
|
|
if (lv_is_active(lv_fast)) {
|
|
log_error("LV %s must be inactive to attach.", display_lvname(lv_fast));
|
|
goto bad;
|
|
}
|
|
|
|
if (!arg_is_set(cmd, yes_ARG) &&
|
|
yes_no_prompt("Erase all existing data on %s? [y/n]: ", display_lvname(lv_fast)) == 'n') {
|
|
log_error("Conversion aborted.");
|
|
goto bad;
|
|
}
|
|
} else {
|
|
if (!_lv_create_cachevol(cmd, vg, lv, &lv_fast))
|
|
goto_bad;
|
|
}
|
|
|
|
is_active = lv_is_active(lv);
|
|
|
|
if (!get_writecache_settings(cmd, &settings, &block_size_sectors)) {
|
|
log_error("Invalid writecache settings.");
|
|
goto bad;
|
|
}
|
|
|
|
if (!is_active) {
|
|
/* checking block size of fs on the lv requires the lv to be active */
|
|
if (!activate_lv(cmd, lv)) {
|
|
log_error("Failed to activate LV to check block size %s", display_lvname(lv));
|
|
goto bad;
|
|
}
|
|
if (!sync_local_dev_names(cmd)) {
|
|
log_error("Failed to sync local dev names.");
|
|
if (!deactivate_lv(cmd, lv))
|
|
stack;
|
|
goto bad;
|
|
}
|
|
}
|
|
|
|
if (!_set_writecache_block_size(cmd, lv, &block_size_sectors)) {
|
|
if (!is_active && !deactivate_lv(cmd, lv))
|
|
stack;
|
|
goto_bad;
|
|
}
|
|
|
|
if (!_check_writecache_memory(cmd, lv_fast, block_size_sectors)) {
|
|
if (!is_active && !deactivate_lv(cmd, lv))
|
|
stack;
|
|
goto_bad;
|
|
}
|
|
|
|
if (!is_active) {
|
|
if (!deactivate_lv(cmd, lv)) {
|
|
log_error("Failed to deactivate LV after checking block size %s", display_lvname(lv));
|
|
goto bad;
|
|
}
|
|
}
|
|
|
|
/* Ensure the LV is not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
goto_bad;
|
|
if (fast_name && !lockd_lv(cmd, lv_fast, "ex", 0))
|
|
goto_bad;
|
|
|
|
/*
|
|
* lv keeps the same lockd lock it had before, the lock for
|
|
* lv_fast is kept but is not used while it's attached, and
|
|
* lv_wcorig gets no lock.
|
|
*/
|
|
if (vg_is_shared(vg) && lv_fast->lock_args) {
|
|
lockd_fast_args = dm_pool_strdup(lv_fast->vg->vgmem, lv_fast->lock_args);
|
|
lockd_fast_name = dm_pool_strdup(lv_fast->vg->vgmem, lv_fast->name);
|
|
lockd_fast_id = lv_fast->lvid.id[1];
|
|
}
|
|
|
|
if (!_writecache_zero(cmd, lv_fast)) {
|
|
log_error("LV %s could not be zeroed.", display_lvname(lv_fast));
|
|
goto bad;
|
|
}
|
|
|
|
/*
|
|
* The lvm tradition is to rename an LV with a special role-specific
|
|
* suffix when it becomes hidden. Here the _cvol suffix is added to
|
|
* the fast LV name. When the cache is detached, it's renamed back.
|
|
*/
|
|
if (dm_snprintf(cvol_name, sizeof(cvol_name), "%s_cvol", lv_fast->name) < 0) {
|
|
log_error("Can't prepare new metadata name for %s.", display_lvname(lv_fast));
|
|
goto bad;
|
|
}
|
|
if (!lv_rename_update(cmd, lv_fast, cvol_name, 0))
|
|
goto_bad;
|
|
|
|
lv_fast->status |= LV_CACHE_VOL;
|
|
|
|
/* When the lv arg is a thinpool, redirect update to data sub lv. */
|
|
|
|
if (lv_is_thin_pool(lv)) {
|
|
lv_update = seg_lv(first_seg(lv), 0);
|
|
log_verbose("Redirecting operation to data sub LV %s.", display_lvname(lv_update));
|
|
} else {
|
|
lv_update = lv;
|
|
}
|
|
|
|
/*
|
|
* Changes the vg struct to match the desired state.
|
|
*
|
|
* - lv keeps existing lv name and id, gets new segment with segtype
|
|
* "writecache".
|
|
*
|
|
* - lv_fast keeps its existing name and id, becomes hidden.
|
|
*
|
|
* - lv_wcorig gets new name (existing name + _wcorig suffix),
|
|
* gets new id, becomes hidden, gets segments from lv.
|
|
*/
|
|
|
|
if (!(lv_wcorig = _lv_writecache_create(cmd, lv_update, lv_fast, block_size_sectors, &settings)))
|
|
goto_bad;
|
|
|
|
/*
|
|
* vg_write(), suspend_lv(), vg_commit(), resume_lv(),
|
|
* where the old LV is suspended and the new LV is resumed.
|
|
*/
|
|
|
|
if (!lv_update_and_reload(lv_update))
|
|
goto_bad;
|
|
|
|
lockd_lv(cmd, lv, "un", 0);
|
|
|
|
if (lockd_fast_name) {
|
|
/* lockd unlock for lv_fast */
|
|
if (!lockd_lv_name(cmd, vg, lockd_fast_name, &lockd_fast_id, lockd_fast_args, "un", 0))
|
|
log_error("Failed to unlock fast LV %s/%s", vg->name, lockd_fast_name);
|
|
}
|
|
|
|
log_print_unless_silent("Logical volume %s now has writecache.",
|
|
display_lvname(lv));
|
|
return ECMD_PROCESSED;
|
|
bad:
|
|
return ECMD_FAILED;
|
|
|
|
}
|
|
|
|
int lvconvert_to_writecache_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_result lr = { 0 };
|
|
int ret;
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lr;
|
|
|
|
cmd->get_vgname_from_options = 0;
|
|
|
|
ret = process_each_lv(cmd, cmd->position_argc, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE, handle, NULL,
|
|
&lvconvert_writecache_attach_single);
|
|
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
int lvconvert_to_cache_with_cachevol_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
struct lvconvert_result lr = { 0 };
|
|
int ret;
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
handle->custom_handle = &lr;
|
|
|
|
cmd->get_vgname_from_options = 0;
|
|
|
|
ret = process_each_lv(cmd, cmd->position_argc, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE, handle, NULL,
|
|
&lvconvert_cachevol_attach_single);
|
|
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
static int _lvconvert_integrity_remove(struct cmd_context *cmd, struct logical_volume *lv)
|
|
{
|
|
if (!lv_is_integrity(lv) && !lv_is_raid(lv)) {
|
|
log_error("LV %s does not have integrity.", display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (!lv_is_raid(lv)) {
|
|
log_error("Cannot remove integrity from non raid type LV %s.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
/* ensure it's not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
return_0;
|
|
|
|
if (!lv_remove_integrity_from_raid(lv))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Logical volume %s has removed integrity.", display_lvname(lv));
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_integrity_add(struct cmd_context *cmd, struct logical_volume *lv,
|
|
struct integrity_settings *set)
|
|
{
|
|
struct volume_group *vg = lv->vg;
|
|
struct dm_list *use_pvh;
|
|
|
|
/* ensure it's not active elsewhere. */
|
|
if (!lockd_lv(cmd, lv, "ex", 0))
|
|
return_0;
|
|
|
|
if (cmd->position_argc > 1) {
|
|
/* First pos arg is required LV, remaining are optional PVs. */
|
|
if (!(use_pvh = create_pv_list(cmd->mem, vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
|
|
return_0;
|
|
} else
|
|
use_pvh = &vg->pvs;
|
|
|
|
if (lv_is_partial(lv)) {
|
|
log_error("Cannot add integrity while LV is missing PVs.");
|
|
return 0;
|
|
}
|
|
|
|
if (!lv_is_raid(lv)) {
|
|
log_error("Cannot add integrity to non raid type LV %s.",
|
|
display_lvname(lv));
|
|
return 0;
|
|
}
|
|
|
|
if (!lv_add_integrity_to_raid(lv, set, use_pvh, NULL))
|
|
return_0;
|
|
|
|
log_print_unless_silent("Logical volume %s has added integrity.", display_lvname(lv));
|
|
return 1;
|
|
}
|
|
|
|
static int _lvconvert_integrity_single(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
struct processing_handle *handle)
|
|
{
|
|
struct integrity_settings settings = { .tag_size = 0 };
|
|
int ret;
|
|
|
|
if (!integrity_mode_set(arg_str_value(cmd, raidintegritymode_ARG, NULL), &settings))
|
|
return_ECMD_FAILED;
|
|
|
|
if (arg_is_set(cmd, raidintegrityblocksize_ARG))
|
|
settings.block_size = arg_int_value(cmd, raidintegrityblocksize_ARG, 0);
|
|
|
|
if (arg_int_value(cmd, raidintegrity_ARG, 0))
|
|
ret = _lvconvert_integrity_add(cmd, lv, &settings);
|
|
else
|
|
ret = _lvconvert_integrity_remove(cmd, lv);
|
|
|
|
if (!ret)
|
|
return_ECMD_FAILED;
|
|
|
|
return ECMD_PROCESSED;
|
|
}
|
|
|
|
int lvconvert_integrity_cmd(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
struct processing_handle *handle;
|
|
int ret;
|
|
|
|
if (!(handle = init_processing_handle(cmd, NULL))) {
|
|
log_error("Failed to initialize processing handle.");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
/* Want to be able to remove integrity from partial LV */
|
|
cmd->handles_missing_pvs = 1;
|
|
|
|
cmd->get_vgname_from_options = 0;
|
|
|
|
ret = process_each_lv(cmd, cmd->position_argc, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE, handle, NULL,
|
|
&_lvconvert_integrity_single);
|
|
|
|
destroy_processing_handle(cmd, handle);
|
|
|
|
return ret;
|
|
}
|
|
|
|
/*
|
|
* All lvconvert command defs have their own function,
|
|
* so the generic function name is unused.
|
|
*/
|
|
|
|
int lvconvert(struct cmd_context *cmd, int argc, char **argv)
|
|
{
|
|
log_error(INTERNAL_ERROR "Missing function for command definition %d:%s.",
|
|
cmd->command->command_index, cmd->command->command_id);
|
|
return ECMD_FAILED;
|
|
}
|