1
0
mirror of git://sourceware.org/git/lvm2.git synced 2024-12-22 17:35:59 +03:00
lvm2/lib/metadata/raid_manip.c

1993 lines
53 KiB
C
Raw Normal View History

/*
* Copyright (C) 2011-2014 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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#include "lib.h"
#include "archiver.h"
#include "metadata.h"
#include "toolcontext.h"
#include "segtype.h"
#include "display.h"
#include "activate.h"
#include "lv_alloc.h"
#include "lvm-string.h"
static int _lv_is_raid_with_tracking(const struct logical_volume *lv,
struct logical_volume **tracking)
{
uint32_t s;
const struct lv_segment *seg = first_seg(lv);
*tracking = NULL;
if (!(lv->status & RAID))
return 0;
for (s = 0; s < seg->area_count; s++)
if (lv_is_visible(seg_lv(seg, s)) &&
!(seg_lv(seg, s)->status & LVM_WRITE))
*tracking = seg_lv(seg, s);
return *tracking ? 1 : 0;
}
int lv_is_raid_with_tracking(const struct logical_volume *lv)
{
struct logical_volume *tracking;
return _lv_is_raid_with_tracking(lv, &tracking);
}
uint32_t lv_raid_image_count(const struct logical_volume *lv)
{
struct lv_segment *seg = first_seg(lv);
if (!seg_is_raid(seg))
return 1;
return seg->area_count;
}
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
static int _activate_sublv_preserving_excl(struct logical_volume *top_lv,
struct logical_volume *sub_lv)
{
struct cmd_context *cmd = top_lv->vg->cmd;
/* If top RAID was EX, use EX */
if (lv_is_active_exclusive_locally(top_lv)) {
if (!activate_lv_excl_local(cmd, sub_lv))
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
return_0;
} else {
if (!activate_lv(cmd, sub_lv))
return_0;
}
return 1;
}
/*
* _raid_in_sync
* @lv
*
* _raid_in_sync works for all types of RAID segtypes, as well
* as 'mirror' segtype. (This is because 'lv_raid_percent' is
* simply a wrapper around 'lv_mirror_percent'.
*
* Returns: 1 if in-sync, 0 otherwise.
*/
2011-08-19 19:59:15 +04:00
static int _raid_in_sync(struct logical_volume *lv)
{
dm_percent_t sync_percent;
if (!lv_raid_percent(lv, &sync_percent)) {
log_error("Unable to determine sync status of %s/%s.",
lv->vg->name, lv->name);
return 0;
}
if (sync_percent == DM_PERCENT_0) {
/*
2015-05-16 01:19:29 +03:00
* FIXME We repeat the status read here to workaround an
* unresolved kernel bug when we see 0 even though the
* the array is 100% in sync.
* https://bugzilla.redhat.com/1210637
*/
if (!lv_raid_percent(lv, &sync_percent)) {
log_error("Unable to determine sync status of %s/%s.",
lv->vg->name, lv->name);
return 0;
}
if (sync_percent == DM_PERCENT_100)
log_warn("WARNING: Sync status for %s is inconsistent.",
display_lvname(lv));
}
return (sync_percent == DM_PERCENT_100) ? 1 : 0;
}
/*
2011-08-19 19:59:15 +04:00
* _raid_remove_top_layer
* @lv
* @removal_list
*
* Remove top layer of RAID LV in order to convert to linear.
* This function makes no on-disk changes. The residual LVs
* returned in 'removal_list' must be freed by the caller.
*
* Returns: 1 on succes, 0 on failure
*/
2011-08-19 19:59:15 +04:00
static int _raid_remove_top_layer(struct logical_volume *lv,
struct dm_list *removal_list)
{
struct lv_list *lvl_array, *lvl;
struct lv_segment *seg = first_seg(lv);
if (!seg_is_mirrored(seg)) {
log_error(INTERNAL_ERROR
"Unable to remove RAID layer from segment type %s",
lvseg_name(seg));
return 0;
}
if (seg->area_count != 1) {
log_error(INTERNAL_ERROR
"Unable to remove RAID layer when there"
" is more than one sub-lv");
return 0;
}
if (!(lvl_array = dm_pool_alloc(lv->vg->vgmem, 2 * sizeof(*lvl))))
return_0;
/* Add last metadata area to removal_list */
lvl_array[0].lv = seg_metalv(seg, 0);
lv_set_visible(seg_metalv(seg, 0));
if (!remove_seg_from_segs_using_this_lv(seg_metalv(seg, 0), seg))
return_0;
seg_metatype(seg, 0) = AREA_UNASSIGNED;
dm_list_add(removal_list, &(lvl_array[0].list));
/* Remove RAID layer and add residual LV to removal_list*/
seg_lv(seg, 0)->status &= ~RAID_IMAGE;
lv_set_visible(seg_lv(seg, 0));
lvl_array[1].lv = seg_lv(seg, 0);
dm_list_add(removal_list, &(lvl_array[1].list));
if (!remove_layer_from_lv(lv, seg_lv(seg, 0)))
return_0;
lv->status &= ~(MIRRORED | RAID);
return 1;
}
/*
2011-08-19 19:59:15 +04:00
* _clear_lv
* @lv
*
* If LV is active:
* clear first block of device
* otherwise:
* activate, clear, deactivate
*
* Returns: 1 on success, 0 on failure
*/
2011-08-19 19:59:15 +04:00
static int _clear_lv(struct logical_volume *lv)
{
int was_active = lv_is_active_locally(lv);
if (test_mode())
return 1;
lv->status |= LV_TEMPORARY;
if (!was_active && !activate_lv_local(lv->vg->cmd, lv)) {
log_error("Failed to activate localy %s for clearing",
lv->name);
return 0;
}
lv->status &= ~LV_TEMPORARY;
log_verbose("Clearing metadata area of %s/%s",
lv->vg->name, lv->name);
/*
* Rather than wiping lv->size, we can simply
2011-08-19 23:35:50 +04:00
* wipe the first sector to remove the superblock of any previous
* RAID devices. It is much quicker.
*/
if (!wipe_lv(lv, (struct wipe_params) { .do_zero = 1, .zero_sectors = 1 })) {
log_error("Failed to zero %s", lv->name);
return 0;
}
if (!was_active && !deactivate_lv(lv->vg->cmd, lv)) {
log_error("Failed to deactivate %s", lv->name);
return 0;
}
return 1;
}
/* Makes on-disk metadata changes */
2011-08-19 19:59:15 +04:00
static int _clear_lvs(struct dm_list *lv_list)
{
struct lv_list *lvl;
struct volume_group *vg = NULL;
if (dm_list_empty(lv_list)) {
log_debug_metadata(INTERNAL_ERROR "Empty list of LVs given for clearing");
return 1;
}
dm_list_iterate_items(lvl, lv_list) {
if (!lv_is_visible(lvl->lv)) {
log_error(INTERNAL_ERROR
"LVs must be set visible before clearing");
return 0;
}
vg = lvl->lv->vg;
}
/*
* FIXME: only vg_[write|commit] if LVs are not already written
* as visible in the LVM metadata (which is never the case yet).
*/
if (!vg || !vg_write(vg) || !vg_commit(vg))
return_0;
dm_list_iterate_items(lvl, lv_list)
2011-08-19 19:59:15 +04:00
if (!_clear_lv(lvl->lv))
return 0;
return 1;
}
/*
* _shift_and_rename_image_components
* @seg: Top-level RAID segment
*
* Shift all higher indexed segment areas down to fill in gaps where
* there are 'AREA_UNASSIGNED' areas and rename data/metadata LVs so
* that their names match their new index. When finished, set
* seg->area_count to new reduced total.
*
* Returns: 1 on success, 0 on failure
*/
static int _shift_and_rename_image_components(struct lv_segment *seg)
{
int len;
char *shift_name;
uint32_t s, missing;
struct cmd_context *cmd = seg->lv->vg->cmd;
/*
* All LVs must be properly named for their index before
* shifting begins. (e.g. Index '0' must contain *_rimage_0 and
* *_rmeta_0. Index 'n' must contain *_rimage_n and *_rmeta_n.)
*/
if (!seg_is_raid(seg))
return_0;
if (seg->area_count > 10) {
/*
* FIXME: Handling more would mean I'd have
* to handle double digits
*/
log_error("Unable handle arrays with more than 10 devices");
return 0;
}
log_very_verbose("Shifting images in %s", seg->lv->name);
for (s = 0, missing = 0; s < seg->area_count; s++) {
if (seg_type(seg, s) == AREA_UNASSIGNED) {
if (seg_metatype(seg, s) != AREA_UNASSIGNED) {
log_error(INTERNAL_ERROR "Metadata segment area"
" #%d should be AREA_UNASSIGNED", s);
return 0;
}
missing++;
continue;
}
if (!missing)
continue;
log_very_verbose("Shifting %s and %s by %u",
seg_metalv(seg, s)->name,
seg_lv(seg, s)->name, missing);
/* Alter rmeta name */
shift_name = dm_pool_strdup(cmd->mem, seg_metalv(seg, s)->name);
if (!shift_name) {
log_error("Memory allocation failed.");
return 0;
}
len = strlen(shift_name) - 1;
shift_name[len] -= missing;
seg_metalv(seg, s)->name = shift_name;
/* Alter rimage name */
shift_name = dm_pool_strdup(cmd->mem, seg_lv(seg, s)->name);
if (!shift_name) {
log_error("Memory allocation failed.");
return 0;
}
len = strlen(shift_name) - 1;
shift_name[len] -= missing;
seg_lv(seg, s)->name = shift_name;
seg->areas[s - missing] = seg->areas[s];
seg->meta_areas[s - missing] = seg->meta_areas[s];
}
seg->area_count -= missing;
return 1;
}
/* Generate raid subvolume name and validate it */
static char *_generate_raid_name(struct logical_volume *lv,
const char *suffix, int count)
{
const char *format = (count >= 0) ? "%s_%s_%u" : "%s_%s";
size_t len = strlen(lv->name) + strlen(suffix) + ((count >= 0) ? 5 : 2);
char *name;
if (!(name = dm_pool_alloc(lv->vg->vgmem, len))) {
log_error("Failed to allocate new name.");
return NULL;
}
if (dm_snprintf(name, len, format, lv->name, suffix, count) < 0)
return_NULL;
if (!validate_name(name)) {
log_error("New logical volume name \"%s\" is not valid.", name);
return NULL;
}
if (find_lv_in_vg(lv->vg, name)) {
log_error("Logical volume %s already exists in volume group %s.",
name, lv->vg->name);
return NULL;
}
return name;
}
/*
* Create an LV of specified type. Set visible after creation.
* This function does not make metadata changes.
*/
static struct logical_volume *_alloc_image_component(struct logical_volume *lv,
const char *alt_base_name,
struct alloc_handle *ah, uint32_t first_area,
uint64_t type)
{
uint64_t status;
char img_name[NAME_LEN];
const char *type_suffix;
struct logical_volume *tmp_lv;
const struct segment_type *segtype;
if (!ah) {
log_error(INTERNAL_ERROR
"Stand-alone %s area allocation not implemented",
(type == RAID_META) ? "metadata" : "data");
return 0;
}
switch (type) {
case RAID_META:
type_suffix = "rmeta";
break;
case RAID_IMAGE:
type_suffix = "rimage";
break;
default:
log_error(INTERNAL_ERROR
"Bad type provided to _alloc_raid_component.");
return 0;
}
if (dm_snprintf(img_name, sizeof(img_name), "%s_%s_%%d",
(alt_base_name) ? : lv->name, type_suffix) < 0) {
log_error("Component name for raid %s is too long.", lv->name);
return 0;
}
status = LVM_READ | LVM_WRITE | LV_REBUILD | type;
if (!(tmp_lv = lv_create_empty(img_name, NULL, status, ALLOC_INHERIT, lv->vg))) {
log_error("Failed to allocate new raid component, %s.", img_name);
return 0;
}
if (!(segtype = get_segtype_from_string(lv->vg->cmd, SEG_TYPE_NAME_STRIPED)))
return_0;
if (!lv_add_segment(ah, first_area, 1, tmp_lv, segtype, 0, status, 0)) {
log_error("Failed to add segment to LV, %s", img_name);
return 0;
}
lv_set_visible(tmp_lv);
return tmp_lv;
}
static int _alloc_image_components(struct logical_volume *lv,
struct dm_list *pvs, uint32_t count,
struct dm_list *new_meta_lvs,
struct dm_list *new_data_lvs)
{
uint32_t s;
uint32_t region_size;
uint32_t extents;
struct lv_segment *seg = first_seg(lv);
const struct segment_type *segtype;
struct alloc_handle *ah;
struct dm_list *parallel_areas;
struct lv_list *lvl_array;
if (!(lvl_array = dm_pool_alloc(lv->vg->vgmem,
sizeof(*lvl_array) * count * 2)))
return_0;
if (!(parallel_areas = build_parallel_areas_from_lv(lv, 0, 1)))
return_0;
if (seg_is_linear(seg))
region_size = get_default_region_size(lv->vg->cmd);
else
region_size = seg->region_size;
if (seg_is_raid(seg))
segtype = seg->segtype;
else if (!(segtype = get_segtype_from_string(lv->vg->cmd, SEG_TYPE_NAME_RAID1)))
return_0;
/*
* The number of extents is based on the RAID type. For RAID1,
* each of the rimages is the same size - 'le_count'. However
* for RAID 4/5/6, the stripes add together (NOT including the parity
* devices) to equal 'le_count'. Thus, when we are allocating
* individual devies, we must specify how large the individual device
* is along with the number we want ('count').
*/
extents = (segtype->parity_devs) ?
(lv->le_count / (seg->area_count - segtype->parity_devs)) :
lv->le_count;
if (!(ah = allocate_extents(lv->vg, NULL, segtype, 0, count, count,
region_size, extents, pvs,
lv->alloc, 0, parallel_areas)))
return_0;
for (s = 0; s < count; ++s) {
/*
* The allocation areas are grouped together. First
* come the rimage allocated areas, then come the metadata
* allocated areas. Thus, the metadata areas are pulled
* from 's + count'.
*/
if (!(lvl_array[s + count].lv =
_alloc_image_component(lv, NULL, ah, s + count, RAID_META))) {
alloc_destroy(ah);
return_0;
}
dm_list_add(new_meta_lvs, &(lvl_array[s + count].list));
if (!(lvl_array[s].lv =
_alloc_image_component(lv, NULL, ah, s, RAID_IMAGE))) {
alloc_destroy(ah);
return_0;
}
dm_list_add(new_data_lvs, &(lvl_array[s].list));
}
alloc_destroy(ah);
return 1;
}
/*
* _alloc_rmeta_for_lv
* @lv
*
* Allocate a RAID metadata device for the given LV (which is or will
* be the associated RAID data device). The new metadata device must
* be allocated from the same PV(s) as the data device.
*/
static int _alloc_rmeta_for_lv(struct logical_volume *data_lv,
struct logical_volume **meta_lv)
{
struct dm_list allocatable_pvs;
struct alloc_handle *ah;
struct lv_segment *seg = first_seg(data_lv);
char *p, base_name[NAME_LEN];
dm_list_init(&allocatable_pvs);
if (!seg_is_linear(seg)) {
log_error(INTERNAL_ERROR "Unable to allocate RAID metadata "
"area for non-linear LV, %s", data_lv->name);
return 0;
}
(void) dm_strncpy(base_name, data_lv->name, sizeof(base_name));
if ((p = strstr(base_name, "_mimage_")))
*p = '\0';
if (!get_pv_list_for_lv(data_lv->vg->cmd->mem,
data_lv, &allocatable_pvs)) {
log_error("Failed to build list of PVs for %s/%s",
data_lv->vg->name, data_lv->name);
return 0;
}
if (!(ah = allocate_extents(data_lv->vg, NULL, seg->segtype, 0, 1, 0,
seg->region_size,
1 /*RAID_METADATA_AREA_LEN*/,
&allocatable_pvs, data_lv->alloc, 0, NULL)))
return_0;
if (!(*meta_lv = _alloc_image_component(data_lv, base_name, ah, 0, RAID_META))) {
alloc_destroy(ah);
return_0;
}
alloc_destroy(ah);
return 1;
}
2011-08-19 19:59:15 +04:00
static int _raid_add_images(struct logical_volume *lv,
uint32_t new_count, struct dm_list *pvs)
{
int rebuild_flag_cleared = 0;
uint32_t s;
uint32_t old_count = lv_raid_image_count(lv);
uint32_t count = new_count - old_count;
uint64_t status_mask = -1;
struct lv_segment *seg = first_seg(lv);
struct dm_list meta_lvs, data_lvs;
struct lv_list *lvl;
struct lv_segment_area *new_areas;
if (lv->status & LV_NOTSYNCED) {
log_error("Can't add image to out-of-sync RAID LV:"
" use 'lvchange --resync' first.");
return 0;
}
if (!_raid_in_sync(lv)) {
log_error("Can't add image to RAID LV that"
" is still initializing.");
return 0;
}
if (!archive(lv->vg))
return_0;
dm_list_init(&meta_lvs); /* For image addition */
dm_list_init(&data_lvs); /* For image addition */
/*
* If the segtype is linear, then we must allocate a metadata
* LV to accompany it.
*/
if (seg_is_linear(seg)) {
/* A complete resync will be done, no need to mark each sub-lv */
status_mask = ~(LV_REBUILD);
if (!(lvl = dm_pool_alloc(lv->vg->vgmem, sizeof(*lvl)))) {
log_error("Memory allocation failed");
return 0;
}
if (!_alloc_rmeta_for_lv(lv, &lvl->lv))
return_0;
dm_list_add(&meta_lvs, &lvl->list);
} else if (!seg_is_raid(seg)) {
log_error("Unable to add RAID images to %s of segment type %s",
lv->name, lvseg_name(seg));
return 0;
}
if (!_alloc_image_components(lv, pvs, count, &meta_lvs, &data_lvs))
return_0;
/*
* If linear, we must correct data LV names. They are off-by-one
* because the linear volume hasn't taken its proper name of "_rimage_0"
* yet. This action must be done before '_clear_lvs' because it
* commits the LVM metadata before clearing the LVs.
*/
if (seg_is_linear(seg)) {
struct dm_list *l;
struct lv_list *lvl_tmp;
dm_list_iterate(l, &data_lvs) {
if (l == dm_list_last(&data_lvs)) {
lvl = dm_list_item(l, struct lv_list);
if (!(lvl->lv->name = _generate_raid_name(lv, "rimage", count)))
return_0;
continue;
}
lvl = dm_list_item(l, struct lv_list);
lvl_tmp = dm_list_item(l->n, struct lv_list);
lvl->lv->name = lvl_tmp->lv->name;
}
}
/* Metadata LVs must be cleared before being added to the array */
2011-08-19 19:59:15 +04:00
if (!_clear_lvs(&meta_lvs))
goto fail;
if (seg_is_linear(seg)) {
first_seg(lv)->status |= RAID_IMAGE;
if (!insert_layer_for_lv(lv->vg->cmd, lv,
RAID | LVM_READ | LVM_WRITE,
"_rimage_0"))
return_0;
lv->status |= RAID;
seg = first_seg(lv);
seg_lv(seg, 0)->status |= RAID_IMAGE | LVM_READ | LVM_WRITE;
seg->region_size = get_default_region_size(lv->vg->cmd);
/* MD's bitmap is limited to tracking 2^21 regions */
while (seg->region_size < (lv->size / (1 << 21))) {
seg->region_size *= 2;
log_very_verbose("Setting RAID1 region_size to %uS",
seg->region_size);
}
if (!(seg->segtype = get_segtype_from_string(lv->vg->cmd, SEG_TYPE_NAME_RAID1)))
return_0;
}
/*
FIXME: It would be proper to activate the new LVs here, instead of having
them activated by the suspend. However, this causes residual device nodes
to be left for these sub-lvs.
dm_list_iterate_items(lvl, &meta_lvs)
if (!do_correct_activate(lv, lvl->lv))
return_0;
dm_list_iterate_items(lvl, &data_lvs)
if (!do_correct_activate(lv, lvl->lv))
return_0;
*/
/* Expand areas array */
if (!(new_areas = dm_pool_zalloc(lv->vg->cmd->mem,
new_count * sizeof(*new_areas)))) {
log_error("Allocation of new areas failed.");
goto fail;
}
memcpy(new_areas, seg->areas, seg->area_count * sizeof(*seg->areas));
seg->areas = new_areas;
/* Expand meta_areas array */
if (!(new_areas = dm_pool_zalloc(lv->vg->cmd->mem,
new_count * sizeof(*new_areas)))) {
log_error("Allocation of new meta areas failed.");
goto fail;
}
if (seg->meta_areas)
memcpy(new_areas, seg->meta_areas,
seg->area_count * sizeof(*seg->meta_areas));
seg->meta_areas = new_areas;
seg->area_count = new_count;
/* Add extra meta area when converting from linear */
s = (old_count == 1) ? 0 : old_count;
/* Set segment areas for metadata sub_lvs */
dm_list_iterate_items(lvl, &meta_lvs) {
log_debug_metadata("Adding %s to %s",
lvl->lv->name, lv->name);
lvl->lv->status &= status_mask;
first_seg(lvl->lv)->status &= status_mask;
if (!set_lv_segment_area_lv(seg, s, lvl->lv, 0,
lvl->lv->status)) {
log_error("Failed to add %s to %s",
lvl->lv->name, lv->name);
goto fail;
}
s++;
}
s = old_count;
/* Set segment areas for data sub_lvs */
dm_list_iterate_items(lvl, &data_lvs) {
log_debug_metadata("Adding %s to %s",
lvl->lv->name, lv->name);
lvl->lv->status &= status_mask;
first_seg(lvl->lv)->status &= status_mask;
if (!set_lv_segment_area_lv(seg, s, lvl->lv, 0,
lvl->lv->status)) {
log_error("Failed to add %s to %s",
lvl->lv->name, lv->name);
goto fail;
}
s++;
}
/*
* FIXME: Failure handling during these points is harder.
*/
dm_list_iterate_items(lvl, &meta_lvs)
lv_set_hidden(lvl->lv);
dm_list_iterate_items(lvl, &data_lvs)
lv_set_hidden(lvl->lv);
if (!lv_update_and_reload_origin(lv))
return_0;
/*
* Now that the 'REBUILD' has made its way to the kernel, we must
* remove the flag so that the individual devices are not rebuilt
* upon every activation.
*/
seg = first_seg(lv);
for (s = 0; s < seg->area_count; s++) {
if ((seg_lv(seg, s)->status & LV_REBUILD) ||
(seg_metalv(seg, s)->status & LV_REBUILD)) {
seg_metalv(seg, s)->status &= ~LV_REBUILD;
seg_lv(seg, s)->status &= ~LV_REBUILD;
rebuild_flag_cleared = 1;
}
}
if (rebuild_flag_cleared) {
if (!vg_write(lv->vg) || !vg_commit(lv->vg)) {
log_error("Failed to clear REBUILD flag for %s/%s components",
lv->vg->name, lv->name);
return 0;
}
backup(lv->vg);
}
return 1;
fail:
2011-08-19 23:35:50 +04:00
/* Cleanly remove newly-allocated LVs that failed insertion attempt */
dm_list_iterate_items(lvl, &meta_lvs)
if (!lv_remove(lvl->lv))
return_0;
dm_list_iterate_items(lvl, &data_lvs)
if (!lv_remove(lvl->lv))
return_0;
return 0;
}
/*
* _extract_image_components
* @seg
* @idx: The index in the areas array to remove
* @extracted_rmeta: The displaced metadata LV
* @extracted_rimage: The displaced data LV
*
* This function extracts the image components - setting the respective
* 'extracted' pointers. It appends '_extracted' to the LVs' names, so that
* there are not future conflicts. It does /not/ commit the results.
* (IOW, erroring-out requires no unwinding of operations.)
*
* This function does /not/ attempt to:
* 1) shift the 'areas' or 'meta_areas' arrays.
* The '[meta_]areas' are left as AREA_UNASSIGNED.
* 2) Adjust the seg->area_count
* 3) Name the extracted LVs appropriately (appends '_extracted' to names)
* These actions must be performed by the caller.
*
* Returns: 1 on success, 0 on failure
*/
static int _extract_image_components(struct lv_segment *seg, uint32_t idx,
struct logical_volume **extracted_rmeta,
struct logical_volume **extracted_rimage)
{
struct logical_volume *data_lv = seg_lv(seg, idx);
struct logical_volume *meta_lv = seg_metalv(seg, idx);
log_very_verbose("Extracting image components %s and %s from %s",
data_lv->name, meta_lv->name, seg->lv->name);
data_lv->status &= ~RAID_IMAGE;
meta_lv->status &= ~RAID_META;
lv_set_visible(data_lv);
lv_set_visible(meta_lv);
/* release removes data and meta areas */
if (!remove_seg_from_segs_using_this_lv(data_lv, seg) ||
!remove_seg_from_segs_using_this_lv(meta_lv, seg))
return_0;
seg_type(seg, idx) = AREA_UNASSIGNED;
seg_metatype(seg, idx) = AREA_UNASSIGNED;
if (!(data_lv->name = _generate_raid_name(data_lv, "_extracted", -1)))
return_0;
if (!(meta_lv->name = _generate_raid_name(meta_lv, "_extracted", -1)))
return_0;
*extracted_rmeta = meta_lv;
*extracted_rimage = data_lv;
return 1;
}
/*
2011-08-19 19:59:15 +04:00
* _raid_extract_images
* @lv
* @new_count: The absolute count of images (e.g. '2' for a 2-way mirror)
* @target_pvs: The list of PVs that are candidates for removal
* @shift: If set, use _shift_and_rename_image_components().
* Otherwise, leave the [meta_]areas as AREA_UNASSIGNED and
* seg->area_count unchanged.
* @extracted_[meta|data]_lvs: The LVs removed from the array. If 'shift'
* is set, then there will likely be name conflicts.
*
* This function extracts _both_ portions of the indexed image. It
* does /not/ commit the results. (IOW, erroring-out requires no unwinding
* of operations.)
*
* Returns: 1 on success, 0 on failure
*/
2011-08-19 19:59:15 +04:00
static int _raid_extract_images(struct logical_volume *lv, uint32_t new_count,
struct dm_list *target_pvs, int shift,
struct dm_list *extracted_meta_lvs,
struct dm_list *extracted_data_lvs)
{
int ss, s, extract, lvl_idx = 0;
struct lv_list *lvl_array;
struct lv_segment *seg = first_seg(lv);
struct logical_volume *rmeta_lv, *rimage_lv;
struct segment_type *error_segtype;
extract = seg->area_count - new_count;
log_verbose("Extracting %u %s from %s/%s", extract,
(extract > 1) ? "images" : "image",
lv->vg->name, lv->name);
if (dm_list_size(target_pvs) < extract) {
log_error("Unable to remove %d images: Only %d device%s given.",
extract, dm_list_size(target_pvs),
(dm_list_size(target_pvs) == 1) ? "" : "s");
return 0;
}
if (!(lvl_array = dm_pool_alloc(lv->vg->vgmem,
sizeof(*lvl_array) * extract * 2)))
return_0;
if (!(error_segtype = get_segtype_from_string(lv->vg->cmd, SEG_TYPE_NAME_ERROR)))
return_0;
/*
* We make two passes over the devices.
* - The first pass we look for error LVs
* - The second pass we look for PVs that match target_pvs
*/
for (ss = (seg->area_count * 2) - 1; (ss >= 0) && extract; ss--) {
s = ss % seg->area_count;
if (ss / seg->area_count) {
/* Conditions for first pass */
if ((first_seg(seg_lv(seg, s))->segtype != error_segtype) &&
(first_seg(seg_metalv(seg, s))->segtype != error_segtype))
continue;
if (!dm_list_empty(target_pvs) &&
(target_pvs != &lv->vg->pvs)) {
/*
* User has supplied a list of PVs, but we
* cannot honor that list because error LVs
* must come first.
*/
log_error("%s has components with error targets"
2014-09-12 01:32:37 +04:00
" that must be removed first: %s.",
display_lvname(lv),
display_lvname(seg_lv(seg, s)));
log_error("Try removing the PV list and rerun"
" the command.");
return 0;
}
log_debug("LVs with error segments to be removed: %s %s",
2014-09-12 01:32:37 +04:00
display_lvname(seg_metalv(seg, s)),
display_lvname(seg_lv(seg, s)));
} else {
/* Conditions for second pass */
if (!target_pvs || !lv_is_on_pvs(seg_lv(seg, s), target_pvs) ||
RAID: Add writemostly/writebehind support for RAID1 'lvchange' is used to alter a RAID 1 logical volume's write-mostly and write-behind characteristics. The '--writemostly' parameter takes a PV as an argument with an optional trailing character to specify whether to set ('y'), unset ('n'), or toggle ('t') the value. If no trailing character is given, it will set the flag. Synopsis: lvchange [--writemostly <PV>:{t|y|n}] [--writebehind <count>] vg/lv Example: lvchange --writemostly /dev/sdb1:y --writebehind 512 vg/raid1_lv The last character in the 'lv_attr' field is used to show whether a device has the WriteMostly flag set. It is signified with a 'w'. If the device has failed, the 'p'artial flag has priority. Example ("nosync" raid1 with mismatch_cnt and writemostly): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg Rwi---r-m 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-w 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-- 1 linear 4.00m Example (raid1 with mismatch_cnt, writemostly - but failed drive): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg rwi---r-p 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-p 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-p 1 linear 4.00m A new reportable field has been added for writebehind as well. If write-behind has not been set or the LV is not RAID1, the field will be blank. Example (writebehind is set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- 512 [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor-- Example (writebehind is not set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor--
2013-04-15 22:59:46 +04:00
!lv_is_on_pvs(seg_metalv(seg, s), target_pvs))
continue;
if (!_raid_in_sync(lv) &&
(!seg_is_mirrored(seg) || (s == 0))) {
log_error("Unable to extract %sRAID image"
" while RAID array is not in-sync",
seg_is_mirrored(seg) ? "primary " : "");
return 0;
}
}
if (!_extract_image_components(seg, s, &rmeta_lv, &rimage_lv)) {
log_error("Failed to extract %s from %s",
seg_lv(seg, s)->name, lv->name);
return 0;
}
if (shift && !_shift_and_rename_image_components(seg)) {
log_error("Failed to shift and rename image components");
return 0;
}
lvl_array[lvl_idx].lv = rmeta_lv;
lvl_array[lvl_idx + 1].lv = rimage_lv;
dm_list_add(extracted_meta_lvs, &(lvl_array[lvl_idx++].list));
dm_list_add(extracted_data_lvs, &(lvl_array[lvl_idx++].list));
extract--;
}
if (extract) {
log_error("Unable to extract enough images to satisfy request");
return 0;
}
return 1;
}
2011-08-19 19:59:15 +04:00
static int _raid_remove_images(struct logical_volume *lv,
uint32_t new_count, struct dm_list *pvs)
{
struct dm_list removal_list;
struct lv_list *lvl;
if (!archive(lv->vg))
return_0;
dm_list_init(&removal_list);
2011-08-19 19:59:15 +04:00
if (!_raid_extract_images(lv, new_count, pvs, 1,
&removal_list, &removal_list)) {
log_error("Failed to extract images from %s/%s",
lv->vg->name, lv->name);
return 0;
}
/* Convert to linear? */
if (new_count == 1) {
if (!_raid_remove_top_layer(lv, &removal_list)) {
log_error("Failed to remove RAID layer"
" after linear conversion");
return 0;
}
RAID: Add writemostly/writebehind support for RAID1 'lvchange' is used to alter a RAID 1 logical volume's write-mostly and write-behind characteristics. The '--writemostly' parameter takes a PV as an argument with an optional trailing character to specify whether to set ('y'), unset ('n'), or toggle ('t') the value. If no trailing character is given, it will set the flag. Synopsis: lvchange [--writemostly <PV>:{t|y|n}] [--writebehind <count>] vg/lv Example: lvchange --writemostly /dev/sdb1:y --writebehind 512 vg/raid1_lv The last character in the 'lv_attr' field is used to show whether a device has the WriteMostly flag set. It is signified with a 'w'. If the device has failed, the 'p'artial flag has priority. Example ("nosync" raid1 with mismatch_cnt and writemostly): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg Rwi---r-m 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-w 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-- 1 linear 4.00m Example (raid1 with mismatch_cnt, writemostly - but failed drive): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg rwi---r-p 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-p 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-p 1 linear 4.00m A new reportable field has been added for writebehind as well. If write-behind has not been set or the LV is not RAID1, the field will be blank. Example (writebehind is set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- 512 [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor-- Example (writebehind is not set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor--
2013-04-15 22:59:46 +04:00
lv->status &= ~(LV_NOTSYNCED | LV_WRITEMOSTLY);
first_seg(lv)->writebehind = 0;
}
if (!vg_write(lv->vg)) {
log_error("Failed to write changes to %s in %s",
lv->name, lv->vg->name);
return 0;
}
if (!suspend_lv(lv->vg->cmd, lv)) {
log_error("Failed to suspend %s/%s before committing changes",
lv->vg->name, lv->name);
vg_revert(lv->vg);
return 0;
}
if (!vg_commit(lv->vg)) {
log_error("Failed to commit changes to %s in %s",
lv->name, lv->vg->name);
return 0;
}
/*
* We activate the extracted sub-LVs first so they are renamed
* and won't conflict with the remaining (possibly shifted)
* sub-LVs.
*/
dm_list_iterate_items(lvl, &removal_list) {
if (!activate_lv_excl_local(lv->vg->cmd, lvl->lv)) {
log_error("Failed to resume extracted LVs");
return 0;
}
}
if (!resume_lv(lv->vg->cmd, lv)) {
log_error("Failed to resume %s/%s after committing changes",
lv->vg->name, lv->name);
return 0;
}
if (!sync_local_dev_names(lv->vg->cmd)) {
log_error("Failed to sync local devices after committing changes for %s.",
display_lvname(lv));
return 0;
}
/*
* Eliminate the extracted LVs
*/
if (!dm_list_empty(&removal_list)) {
dm_list_iterate_items(lvl, &removal_list) {
if (!deactivate_lv(lv->vg->cmd, lvl->lv))
return_0;
if (!lv_remove(lvl->lv))
return_0;
}
if (!vg_write(lv->vg) || !vg_commit(lv->vg))
return_0;
}
backup(lv->vg);
return 1;
}
/*
* lv_raid_change_image_count
* @lv
* @new_count: The absolute count of images (e.g. '2' for a 2-way mirror)
* @pvs: The list of PVs that are candidates for removal (or empty list)
*
* RAID arrays have 'images' which are composed of two parts, they are:
* - 'rimage': The data/parity holding portion
* - 'rmeta' : The metadata holding portion (i.e. superblock/bitmap area)
* This function adds or removes _both_ portions of the image and commits
* the results.
*
* Returns: 1 on success, 0 on failure
*/
int lv_raid_change_image_count(struct logical_volume *lv,
uint32_t new_count, struct dm_list *pvs)
{
uint32_t old_count = lv_raid_image_count(lv);
if (old_count == new_count) {
log_warn("%s/%s already has image count of %d.",
lv->vg->name, lv->name, new_count);
return 1;
}
/*
* LV must be either in-active or exclusively active
*/
if (lv_is_active(lv_lock_holder(lv)) && vg_is_clustered(lv->vg) &&
!lv_is_active_exclusive_locally(lv_lock_holder(lv))) {
log_error("%s/%s must be active exclusive locally to"
" perform this operation.", lv->vg->name, lv->name);
return 0;
}
if (old_count > new_count)
2011-08-19 19:59:15 +04:00
return _raid_remove_images(lv, new_count, pvs);
2011-08-19 19:59:15 +04:00
return _raid_add_images(lv, new_count, pvs);
}
int lv_raid_split(struct logical_volume *lv, const char *split_name,
uint32_t new_count, struct dm_list *splittable_pvs)
{
struct lv_list *lvl;
struct dm_list removal_list, data_list;
struct cmd_context *cmd = lv->vg->cmd;
uint32_t old_count = lv_raid_image_count(lv);
struct logical_volume *tracking;
struct dm_list tracking_pvs;
dm_list_init(&removal_list);
dm_list_init(&data_list);
2015-03-05 23:00:44 +03:00
if (is_lockd_type(lv->vg->lock_type)) {
log_error("Splitting raid image is not allowed with lock_type %s",
lv->vg->lock_type);
return 0;
}
if ((old_count - new_count) != 1) {
log_error("Unable to split more than one image from %s/%s",
lv->vg->name, lv->name);
return 0;
}
if (!seg_is_mirrored(first_seg(lv)) ||
seg_is_raid10(first_seg(lv))) {
log_error("Unable to split logical volume of segment type, %s",
lvseg_name(first_seg(lv)));
return 0;
}
if (find_lv_in_vg(lv->vg, split_name)) {
log_error("Logical Volume \"%s\" already exists in %s",
split_name, lv->vg->name);
return 0;
}
2011-08-19 19:59:15 +04:00
if (!_raid_in_sync(lv)) {
log_error("Unable to split %s/%s while it is not in-sync.",
lv->vg->name, lv->name);
return 0;
}
/*
* We only allow a split while there is tracking if it is to
* complete the split of the tracking sub-LV
*/
if (_lv_is_raid_with_tracking(lv, &tracking)) {
RAID: Add writemostly/writebehind support for RAID1 'lvchange' is used to alter a RAID 1 logical volume's write-mostly and write-behind characteristics. The '--writemostly' parameter takes a PV as an argument with an optional trailing character to specify whether to set ('y'), unset ('n'), or toggle ('t') the value. If no trailing character is given, it will set the flag. Synopsis: lvchange [--writemostly <PV>:{t|y|n}] [--writebehind <count>] vg/lv Example: lvchange --writemostly /dev/sdb1:y --writebehind 512 vg/raid1_lv The last character in the 'lv_attr' field is used to show whether a device has the WriteMostly flag set. It is signified with a 'w'. If the device has failed, the 'p'artial flag has priority. Example ("nosync" raid1 with mismatch_cnt and writemostly): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg Rwi---r-m 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-w 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-- 1 linear 4.00m Example (raid1 with mismatch_cnt, writemostly - but failed drive): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg rwi---r-p 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-p 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-p 1 linear 4.00m A new reportable field has been added for writebehind as well. If write-behind has not been set or the LV is not RAID1, the field will be blank. Example (writebehind is set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- 512 [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor-- Example (writebehind is not set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor--
2013-04-15 22:59:46 +04:00
if (!lv_is_on_pvs(tracking, splittable_pvs)) {
log_error("Unable to split additional image from %s "
"while tracking changes for %s",
lv->name, tracking->name);
return 0;
}
/* Ensure we only split the tracking image */
dm_list_init(&tracking_pvs);
splittable_pvs = &tracking_pvs;
if (!get_pv_list_for_lv(tracking->vg->cmd->mem,
tracking, splittable_pvs))
return_0;
}
2011-08-19 19:59:15 +04:00
if (!_raid_extract_images(lv, new_count, splittable_pvs, 1,
&removal_list, &data_list)) {
log_error("Failed to extract images from %s/%s",
lv->vg->name, lv->name);
return 0;
}
/* Convert to linear? */
2011-08-19 19:59:15 +04:00
if ((new_count == 1) && !_raid_remove_top_layer(lv, &removal_list)) {
log_error("Failed to remove RAID layer after linear conversion");
return 0;
}
/* Get first item */
dm_list_iterate_items(lvl, &data_list)
break;
lvl->lv->name = split_name;
if (!vg_write(lv->vg)) {
log_error("Failed to write changes to %s in %s",
lv->name, lv->vg->name);
return 0;
}
if (!suspend_lv(cmd, lv_lock_holder(lv))) {
log_error("Failed to suspend %s/%s before committing changes",
lv->vg->name, lv->name);
vg_revert(lv->vg);
return 0;
}
if (!vg_commit(lv->vg)) {
log_error("Failed to commit changes to %s in %s",
lv->name, lv->vg->name);
return 0;
}
/*
* First activate the newly split LV and LVs on the removal list.
* This is necessary so that there are no name collisions due to
* the original RAID LV having possibly had sub-LVs that have been
* shifted and renamed.
*/
if (!activate_lv_excl_local(cmd, lvl->lv))
return_0;
dm_list_iterate_items(lvl, &removal_list)
if (!activate_lv_excl_local(cmd, lvl->lv))
return_0;
if (!resume_lv(cmd, lv_lock_holder(lv))) {
log_error("Failed to resume %s/%s after committing changes",
lv->vg->name, lv->name);
return 0;
}
/*
* Since newly split LV is typically already active - we need to call
* suspend() and resume() to also rename it.
*
* TODO: activate should recognize it and avoid these 2 calls
*/
/*
* Eliminate the residual LVs
*/
dm_list_iterate_items(lvl, &removal_list) {
if (!deactivate_lv(cmd, lvl->lv))
return_0;
if (!lv_remove(lvl->lv))
return_0;
}
if (!vg_write(lv->vg) || !vg_commit(lv->vg))
return_0;
backup(lv->vg);
return 1;
}
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
/*
* lv_raid_split_and_track
* @lv
* @splittable_pvs
*
* Only allows a single image to be split while tracking. The image
* never actually leaves the mirror. It is simply made visible. This
* action triggers two things: 1) users are able to access the (data) image
* and 2) lower layers replace images marked with a visible flag with
* error targets.
*
* Returns: 1 on success, 0 on error
*/
int lv_raid_split_and_track(struct logical_volume *lv,
struct dm_list *splittable_pvs)
{
int s;
struct lv_segment *seg = first_seg(lv);
if (!seg_is_mirrored(seg)) {
log_error("Unable to split images from non-mirrored RAID");
return 0;
}
2011-08-19 19:59:15 +04:00
if (!_raid_in_sync(lv)) {
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
log_error("Unable to split image from %s/%s while not in-sync",
lv->vg->name, lv->name);
return 0;
}
/* Cannot track two split images at once */
if (lv_is_raid_with_tracking(lv)) {
log_error("Cannot track more than one split image at a time");
return 0;
}
for (s = seg->area_count - 1; s >= 0; --s) {
RAID: Add writemostly/writebehind support for RAID1 'lvchange' is used to alter a RAID 1 logical volume's write-mostly and write-behind characteristics. The '--writemostly' parameter takes a PV as an argument with an optional trailing character to specify whether to set ('y'), unset ('n'), or toggle ('t') the value. If no trailing character is given, it will set the flag. Synopsis: lvchange [--writemostly <PV>:{t|y|n}] [--writebehind <count>] vg/lv Example: lvchange --writemostly /dev/sdb1:y --writebehind 512 vg/raid1_lv The last character in the 'lv_attr' field is used to show whether a device has the WriteMostly flag set. It is signified with a 'w'. If the device has failed, the 'p'artial flag has priority. Example ("nosync" raid1 with mismatch_cnt and writemostly): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg Rwi---r-m 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-w 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-- 1 linear 4.00m Example (raid1 with mismatch_cnt, writemostly - but failed drive): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg rwi---r-p 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-p 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-p 1 linear 4.00m A new reportable field has been added for writebehind as well. If write-behind has not been set or the LV is not RAID1, the field will be blank. Example (writebehind is set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- 512 [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor-- Example (writebehind is not set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor--
2013-04-15 22:59:46 +04:00
if (!lv_is_on_pvs(seg_lv(seg, s), splittable_pvs))
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
continue;
lv_set_visible(seg_lv(seg, s));
seg_lv(seg, s)->status &= ~LVM_WRITE;
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
break;
}
if (s >= (int) seg->area_count) {
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
log_error("Unable to find image to satisfy request");
return 0;
}
if (!lv_update_and_reload(lv))
return_0;
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
log_print_unless_silent("%s split from %s for read-only purposes.",
seg_lv(seg, s)->name, lv->name);
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
/* Activate the split (and tracking) LV */
if (!_activate_sublv_preserving_excl(lv, seg_lv(seg, s)))
return_0;
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
log_print_unless_silent("Use 'lvconvert --merge %s/%s' to merge back into %s",
lv->vg->name, seg_lv(seg, s)->name, lv->name);
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
return 1;
}
int lv_raid_merge(struct logical_volume *image_lv)
{
uint32_t s;
char *p, *lv_name;
struct lv_list *lvl;
struct logical_volume *lv;
struct logical_volume *meta_lv = NULL;
struct lv_segment *seg;
struct volume_group *vg = image_lv->vg;
if (image_lv->status & LVM_WRITE) {
log_error("%s is not read-only - refusing to merge.",
display_lvname(image_lv));
return 0;
}
if (!(lv_name = dm_pool_strdup(vg->vgmem, image_lv->name)))
return_0;
if (!(p = strstr(lv_name, "_rimage_"))) {
2014-09-12 01:32:37 +04:00
log_error("Unable to merge non-mirror image %s.",
display_lvname(image_lv));
return 0;
}
*p = '\0'; /* lv_name is now that of top-level RAID */
if (!(lvl = find_lv_in_vg(vg, lv_name))) {
2014-09-12 01:32:37 +04:00
log_error("Unable to find containing RAID array for %s.",
display_lvname(image_lv));
return 0;
}
lv = lvl->lv;
seg = first_seg(lv);
for (s = 0; s < seg->area_count; ++s)
if (seg_lv(seg, s) == image_lv)
meta_lv = seg_metalv(seg, s);
2014-09-12 01:32:54 +04:00
if (!meta_lv) {
log_error("Failed to find meta for %s in RAID array %s.",
display_lvname(image_lv),
display_lvname(lv));
return 0;
}
if (!deactivate_lv(vg->cmd, meta_lv)) {
2014-09-12 01:32:37 +04:00
log_error("Failed to deactivate %s before merging.",
display_lvname(meta_lv));
return 0;
}
if (!deactivate_lv(vg->cmd, image_lv)) {
2014-09-12 01:32:37 +04:00
log_error("Failed to deactivate %s before merging.",
display_lvname(image_lv));
return 0;
}
lv_set_hidden(image_lv);
image_lv->status |= (lv->status & LVM_WRITE);
image_lv->status |= RAID_IMAGE;
if (!lv_update_and_reload(lv))
return_0;
log_print_unless_silent("%s/%s successfully merged back into %s/%s",
vg->name, image_lv->name, vg->name, lv->name);
return 1;
}
static int _convert_mirror_to_raid1(struct logical_volume *lv,
const struct segment_type *new_segtype)
{
uint32_t s;
struct lv_segment *seg = first_seg(lv);
struct lv_list lvl_array[seg->area_count], *lvl;
struct dm_list meta_lvs;
struct lv_segment_area *meta_areas;
char *new_name;
dm_list_init(&meta_lvs);
if (!_raid_in_sync(lv)) {
log_error("Unable to convert %s/%s while it is not in-sync",
lv->vg->name, lv->name);
return 0;
}
if (!(meta_areas = dm_pool_zalloc(lv->vg->vgmem,
lv_mirror_count(lv) * sizeof(*meta_areas)))) {
log_error("Failed to allocate meta areas memory.");
return 0;
}
if (!archive(lv->vg))
return_0;
for (s = 0; s < seg->area_count; s++) {
log_debug_metadata("Allocating new metadata LV for %s",
seg_lv(seg, s)->name);
if (!_alloc_rmeta_for_lv(seg_lv(seg, s), &(lvl_array[s].lv))) {
log_error("Failed to allocate metadata LV for %s in %s",
seg_lv(seg, s)->name, lv->name);
return 0;
}
dm_list_add(&meta_lvs, &(lvl_array[s].list));
}
log_debug_metadata("Clearing newly allocated metadata LVs");
if (!_clear_lvs(&meta_lvs)) {
log_error("Failed to initialize metadata LVs");
return 0;
}
if (seg->log_lv) {
log_debug_metadata("Removing mirror log, %s", seg->log_lv->name);
if (!remove_mirror_log(lv->vg->cmd, lv, NULL, 0)) {
log_error("Failed to remove mirror log");
return 0;
}
}
seg->meta_areas = meta_areas;
s = 0;
dm_list_iterate_items(lvl, &meta_lvs) {
log_debug_metadata("Adding %s to %s", lvl->lv->name, lv->name);
/* Images are known to be in-sync */
lvl->lv->status &= ~LV_REBUILD;
first_seg(lvl->lv)->status &= ~LV_REBUILD;
lv_set_hidden(lvl->lv);
if (!set_lv_segment_area_lv(seg, s, lvl->lv, 0,
lvl->lv->status)) {
log_error("Failed to add %s to %s",
lvl->lv->name, lv->name);
return 0;
}
s++;
}
for (s = 0; s < seg->area_count; ++s) {
if (!(new_name = _generate_raid_name(lv, "rimage", s)))
return_0;
log_debug_metadata("Renaming %s to %s", seg_lv(seg, s)->name, new_name);
seg_lv(seg, s)->name = new_name;
seg_lv(seg, s)->status &= ~MIRROR_IMAGE;
seg_lv(seg, s)->status |= RAID_IMAGE;
}
init_mirror_in_sync(1);
log_debug_metadata("Setting new segtype for %s", lv->name);
seg->segtype = new_segtype;
lv->status &= ~MIRROR;
lv->status &= ~MIRRORED;
lv->status |= RAID;
seg->status |= RAID;
if (!lv_update_and_reload(lv))
return_0;
return 1;
}
/*
* lv_raid_reshape
* @lv
* @new_segtype
*
* Convert an LV from one RAID type (or 'mirror' segtype) to another.
*
* Returns: 1 on success, 0 on failure
*/
int lv_raid_reshape(struct logical_volume *lv,
const struct segment_type *new_segtype)
{
struct lv_segment *seg = first_seg(lv);
if (!new_segtype) {
log_error(INTERNAL_ERROR "New segtype not specified");
return 0;
}
if (vg_is_clustered(lv->vg) && !lv_is_active_exclusive_locally(lv)) {
log_error("%s/%s must be active exclusive locally to"
" perform this operation.", lv->vg->name, lv->name);
return 0;
}
if (seg_is_mirror(seg) && segtype_is_raid1(new_segtype))
return _convert_mirror_to_raid1(lv, new_segtype);
log_error("Converting the segment type for %s/%s from %s to %s is not yet supported.",
lv->vg->name, lv->name, lvseg_name(seg), new_segtype->name);
return 0;
}
static int _remove_partial_multi_segment_image(struct logical_volume *lv,
struct dm_list *remove_pvs)
{
uint32_t s, extents_needed;
struct lv_segment *rm_seg, *raid_seg = first_seg(lv);
struct logical_volume *rm_image = NULL;
struct physical_volume *pv;
if (!(lv->status & PARTIAL_LV))
return_0;
for (s = 0; s < raid_seg->area_count; s++) {
extents_needed = 0;
if ((seg_lv(raid_seg, s)->status & PARTIAL_LV) &&
lv_is_on_pvs(seg_lv(raid_seg, s), remove_pvs) &&
(dm_list_size(&(seg_lv(raid_seg, s)->segments)) > 1)) {
rm_image = seg_lv(raid_seg, s);
/* First, how many damaged extents are there */
if (seg_metalv(raid_seg, s)->status & PARTIAL_LV)
extents_needed += seg_metalv(raid_seg, s)->le_count;
dm_list_iterate_items(rm_seg, &rm_image->segments) {
/*
* segment areas are for stripe, mirror, raid,
* etc. We only need to check the first area
* if we are dealing with RAID image LVs.
*/
if (seg_type(rm_seg, 0) != AREA_PV)
continue;
pv = seg_pv(rm_seg, 0);
if (pv->status & MISSING_PV)
extents_needed += rm_seg->len;
}
log_debug("%u extents needed to repair %s",
extents_needed, rm_image->name);
/* Second, do the other PVs have the space */
dm_list_iterate_items(rm_seg, &rm_image->segments) {
if (seg_type(rm_seg, 0) != AREA_PV)
continue;
pv = seg_pv(rm_seg, 0);
if (pv->status & MISSING_PV)
continue;
if ((pv->pe_count - pv->pe_alloc_count) >
extents_needed) {
log_debug("%s has enough space for %s",
pv_dev_name(pv),
rm_image->name);
goto has_enough_space;
}
log_debug("Not enough space on %s for %s",
pv_dev_name(pv), rm_image->name);
}
}
}
/*
* This is likely to be the normal case - single
* segment images.
*/
return_0;
has_enough_space:
/*
* Now we have a multi-segment, partial image that has enough
* space on just one of its PVs for the entire image to be
* replaced. So, we replace the image's space with an error
* target so that the allocator can find that space (along with
* the remaining free space) in order to allocate the image
* anew.
*/
if (!replace_lv_with_error_segment(rm_image))
return_0;
return 1;
}
static int _avoid_pvs_of_lv(struct logical_volume *lv, void *data)
{
struct dm_list *allocate_pvs = (struct dm_list *) data;
struct pv_list *pvl;
dm_list_iterate_items(pvl, allocate_pvs)
if (!(lv->status & PARTIAL_LV) &&
lv_is_on_pv(lv, pvl->pv))
pvl->pv->status |= PV_ALLOCATION_PROHIBITED;
return 1;
}
/*
* Prevent any PVs holding other image components of @lv from being used for allocation
* by setting the internal PV_ALLOCATION_PROHIBITED flag to use it to avoid generating
* pv maps for those PVs.
*/
static int _avoid_pvs_with_other_images_of_lv(struct logical_volume *lv, struct dm_list *allocate_pvs)
{
return for_each_sub_lv(lv, _avoid_pvs_of_lv, allocate_pvs);
}
/*
* lv_raid_replace
* @lv
* @remove_pvs
* @allocate_pvs
*
* Replace the specified PVs.
*/
int lv_raid_replace(struct logical_volume *lv,
struct dm_list *remove_pvs,
struct dm_list *allocate_pvs)
{
int partial_segment_removed = 0;
uint32_t s, sd, match_count = 0;
struct dm_list old_lvs;
struct dm_list new_meta_lvs, new_data_lvs;
struct lv_segment *raid_seg = first_seg(lv);
struct lv_list *lvl;
char *tmp_names[raid_seg->area_count * 2];
dm_list_init(&old_lvs);
dm_list_init(&new_meta_lvs);
dm_list_init(&new_data_lvs);
if (lv->status & PARTIAL_LV)
lv->vg->cmd->partial_activation = 1;
if (!lv_is_active_exclusive_locally(lv_lock_holder(lv))) {
log_error("%s/%s must be active %sto perform this operation.",
lv->vg->name, lv->name,
vg_is_clustered(lv->vg) ? "exclusive locally " : "");
return 0;
}
if (!mirror_in_sync() && !_raid_in_sync(lv)) {
log_error("Unable to replace devices in %s/%s while it is"
" not in-sync.", lv->vg->name, lv->name);
return 0;
}
if (!archive(lv->vg))
return_0;
/*
* How many sub-LVs are being removed?
*/
for (s = 0; s < raid_seg->area_count; s++) {
if ((seg_type(raid_seg, s) == AREA_UNASSIGNED) ||
(seg_metatype(raid_seg, s) == AREA_UNASSIGNED)) {
log_error("Unable to replace RAID images while the "
"array has unassigned areas");
return 0;
}
if (lv_is_virtual(seg_lv(raid_seg, s)) ||
lv_is_virtual(seg_metalv(raid_seg, s)) ||
RAID: Add writemostly/writebehind support for RAID1 'lvchange' is used to alter a RAID 1 logical volume's write-mostly and write-behind characteristics. The '--writemostly' parameter takes a PV as an argument with an optional trailing character to specify whether to set ('y'), unset ('n'), or toggle ('t') the value. If no trailing character is given, it will set the flag. Synopsis: lvchange [--writemostly <PV>:{t|y|n}] [--writebehind <count>] vg/lv Example: lvchange --writemostly /dev/sdb1:y --writebehind 512 vg/raid1_lv The last character in the 'lv_attr' field is used to show whether a device has the WriteMostly flag set. It is signified with a 'w'. If the device has failed, the 'p'artial flag has priority. Example ("nosync" raid1 with mismatch_cnt and writemostly): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg Rwi---r-m 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-w 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-- 1 linear 4.00m Example (raid1 with mismatch_cnt, writemostly - but failed drive): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg rwi---r-p 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-p 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-p 1 linear 4.00m A new reportable field has been added for writebehind as well. If write-behind has not been set or the LV is not RAID1, the field will be blank. Example (writebehind is set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- 512 [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor-- Example (writebehind is not set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor--
2013-04-15 22:59:46 +04:00
lv_is_on_pvs(seg_lv(raid_seg, s), remove_pvs) ||
lv_is_on_pvs(seg_metalv(raid_seg, s), remove_pvs))
match_count++;
}
if (!match_count) {
log_verbose("%s/%s does not contain devices specified"
" for replacement", lv->vg->name, lv->name);
return 1;
} else if (match_count == raid_seg->area_count) {
log_error("Unable to remove all PVs from %s/%s at once.",
lv->vg->name, lv->name);
return 0;
} else if (raid_seg->segtype->parity_devs &&
(match_count > raid_seg->segtype->parity_devs)) {
log_error("Unable to replace more than %u PVs from (%s) %s/%s",
raid_seg->segtype->parity_devs,
lvseg_name(raid_seg),
lv->vg->name, lv->name);
return 0;
} else if (seg_is_raid10(raid_seg)) {
uint32_t i, rebuilds_per_group = 0;
/* FIXME: We only support 2-way mirrors in RAID10 currently */
uint32_t copies = 2;
for (i = 0; i < raid_seg->area_count * copies; i++) {
s = i % raid_seg->area_count;
if (!(i % copies))
rebuilds_per_group = 0;
RAID: Add writemostly/writebehind support for RAID1 'lvchange' is used to alter a RAID 1 logical volume's write-mostly and write-behind characteristics. The '--writemostly' parameter takes a PV as an argument with an optional trailing character to specify whether to set ('y'), unset ('n'), or toggle ('t') the value. If no trailing character is given, it will set the flag. Synopsis: lvchange [--writemostly <PV>:{t|y|n}] [--writebehind <count>] vg/lv Example: lvchange --writemostly /dev/sdb1:y --writebehind 512 vg/raid1_lv The last character in the 'lv_attr' field is used to show whether a device has the WriteMostly flag set. It is signified with a 'w'. If the device has failed, the 'p'artial flag has priority. Example ("nosync" raid1 with mismatch_cnt and writemostly): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg Rwi---r-m 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-w 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-- 1 linear 4.00m Example (raid1 with mismatch_cnt, writemostly - but failed drive): [~]# lvs -a --segment vg LV VG Attr #Str Type SSize raid1 vg rwi---r-p 2 raid1 500.00m [raid1_rimage_0] vg Iwi---r-- 1 linear 500.00m [raid1_rimage_1] vg Iwi---r-p 1 linear 500.00m [raid1_rmeta_0] vg ewi---r-- 1 linear 4.00m [raid1_rmeta_1] vg ewi---r-p 1 linear 4.00m A new reportable field has been added for writebehind as well. If write-behind has not been set or the LV is not RAID1, the field will be blank. Example (writebehind is set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- 512 [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor-- Example (writebehind is not set): [~]# lvs -a -o name,attr,writebehind vg LV Attr WBehind lv rwi-a-r-- [lv_rimage_0] iwi-aor-w [lv_rimage_1] iwi-aor-- [lv_rmeta_0] ewi-aor-- [lv_rmeta_1] ewi-aor--
2013-04-15 22:59:46 +04:00
if (lv_is_on_pvs(seg_lv(raid_seg, s), remove_pvs) ||
lv_is_on_pvs(seg_metalv(raid_seg, s), remove_pvs) ||
lv_is_virtual(seg_lv(raid_seg, s)) ||
lv_is_virtual(seg_metalv(raid_seg, s)))
rebuilds_per_group++;
if (rebuilds_per_group >= copies) {
log_error("Unable to replace all the devices "
"in a RAID10 mirror group.");
return 0;
}
}
}
/* Prevent any PVs holding image components from being used for allocation */
if (!_avoid_pvs_with_other_images_of_lv(lv, allocate_pvs)) {
log_error("Failed to prevent PVs holding image components "
"from being used for allocation.");
return 0;
}
/*
* Allocate the new image components first
* - This makes it easy to avoid all currently used devs
* - We can immediately tell if there is enough space
*
* - We need to change the LV names when we insert them.
*/
try_again:
if (!_alloc_image_components(lv, allocate_pvs, match_count,
&new_meta_lvs, &new_data_lvs)) {
2014-09-12 01:32:54 +04:00
if (!(lv->status & PARTIAL_LV)) {
log_error("LV %s in not partial.", display_lvname(lv));
return 0;
2014-09-12 01:32:54 +04:00
}
/* This is a repair, so try to do better than all-or-nothing */
match_count--;
if (match_count > 0) {
log_error("Failed to replace %u devices."
" Attempting to replace %u instead.",
match_count, match_count+1);
/*
* Since we are replacing some but not all of the bad
* devices, we must set partial_activation
*/
lv->vg->cmd->partial_activation = 1;
goto try_again;
} else if (!match_count && !partial_segment_removed) {
/*
* We are down to the last straw. We can only hope
* that a failed PV is just one of several PVs in
* the image; and if we extract the image, there may
* be enough room on the image's other PVs for a
* reallocation of the image.
*/
if (!_remove_partial_multi_segment_image(lv, remove_pvs))
return_0;
match_count = 1;
partial_segment_removed = 1;
lv->vg->cmd->partial_activation = 1;
goto try_again;
}
log_error("Failed to allocate replacement images for %s/%s",
lv->vg->name, lv->name);
return 0;
}
/*
* Remove the old images
* - If we did this before the allocate, we wouldn't have to rename
* the allocated images, but it'd be much harder to avoid the right
* PVs during allocation.
*
* - If this is a repair and we were forced to call
* _remove_partial_multi_segment_image, then the remove_pvs list
* is no longer relevant - _raid_extract_images is forced to replace
* the image with the error target. Thus, the full set of PVs is
* supplied - knowing that only the image with the error target
* will be affected.
*/
if (!_raid_extract_images(lv, raid_seg->area_count - match_count,
partial_segment_removed ?
&lv->vg->pvs : remove_pvs, 0,
&old_lvs, &old_lvs)) {
log_error("Failed to remove the specified images from %s/%s",
lv->vg->name, lv->name);
return 0;
}
/*
* Now that they are extracted and visible, make the system aware
* of their new names.
*/
dm_list_iterate_items(lvl, &old_lvs)
if (!activate_lv_excl_local(lv->vg->cmd, lvl->lv))
return_0;
/*
* Skip metadata operation normally done to clear the metadata sub-LVs.
*
* The LV_REBUILD flag is set on the new sub-LVs,
* so they will be rebuilt and we don't need to clear the metadata dev.
*/
for (s = 0; s < raid_seg->area_count; s++) {
sd = s + raid_seg->area_count;
if ((seg_type(raid_seg, s) == AREA_UNASSIGNED) &&
(seg_metatype(raid_seg, s) == AREA_UNASSIGNED)) {
/* Adjust the new metadata LV name */
lvl = dm_list_item(dm_list_first(&new_meta_lvs),
struct lv_list);
dm_list_del(&lvl->list);
if (!(tmp_names[s] = _generate_raid_name(lv, "rmeta", s)))
return_0;
if (!set_lv_segment_area_lv(raid_seg, s, lvl->lv, 0,
lvl->lv->status)) {
log_error("Failed to add %s to %s",
lvl->lv->name, lv->name);
return 0;
}
lv_set_hidden(lvl->lv);
/* Adjust the new data LV name */
lvl = dm_list_item(dm_list_first(&new_data_lvs),
struct lv_list);
dm_list_del(&lvl->list);
if (!(tmp_names[sd] = _generate_raid_name(lv, "rimage", s)))
return_0;
if (!set_lv_segment_area_lv(raid_seg, s, lvl->lv, 0,
lvl->lv->status)) {
log_error("Failed to add %s to %s",
lvl->lv->name, lv->name);
return 0;
}
lv_set_hidden(lvl->lv);
} else
tmp_names[s] = tmp_names[sd] = NULL;
}
if (!lv_update_and_reload_origin(lv))
return_0;
dm_list_iterate_items(lvl, &old_lvs) {
if (!deactivate_lv(lv->vg->cmd, lvl->lv))
return_0;
if (!lv_remove(lvl->lv))
return_0;
}
/* Update new sub-LVs to correct name and clear REBUILD flag */
for (s = 0; s < raid_seg->area_count; s++) {
sd = s + raid_seg->area_count;
if (tmp_names[s] && tmp_names[sd]) {
seg_metalv(raid_seg, s)->name = tmp_names[s];
seg_lv(raid_seg, s)->name = tmp_names[sd];
seg_metalv(raid_seg, s)->status &= ~LV_REBUILD;
seg_lv(raid_seg, s)->status &= ~LV_REBUILD;
}
}
if (!lv_update_and_reload_origin(lv))
return_0;
return 1;
}
int lv_raid_remove_missing(struct logical_volume *lv)
{
2013-04-11 15:48:23 +04:00
uint32_t s;
struct lv_segment *seg = first_seg(lv);
if (!(lv->status & PARTIAL_LV)) {
log_error(INTERNAL_ERROR "%s/%s is not a partial LV",
lv->vg->name, lv->name);
return 0;
}
if (!archive(lv->vg))
return_0;
log_debug("Attempting to remove missing devices from %s LV, %s",
lvseg_name(seg), lv->name);
/*
* FIXME: Make sure # of compromised components will not affect RAID
*/
2013-04-11 15:48:23 +04:00
for (s = 0; s < seg->area_count; s++) {
if (!(seg_lv(seg, s)->status & PARTIAL_LV) &&
!(seg_metalv(seg, s)->status & PARTIAL_LV))
continue;
log_debug("Replacing %s and %s segments with error target",
seg_lv(seg, s)->name, seg_metalv(seg, s)->name);
if (!replace_lv_with_error_segment(seg_lv(seg, s))) {
2014-09-12 01:32:37 +04:00
log_error("Failed to replace %s's extents with error target.",
display_lvname(seg_lv(seg, s)));
return 0;
}
if (!replace_lv_with_error_segment(seg_metalv(seg, s))) {
2014-09-12 01:32:37 +04:00
log_error("Failed to replace %s's extents with error target.",
display_lvname(seg_metalv(seg, s)));
return 0;
}
}
if (!lv_update_and_reload(lv))
return_0;
return 1;
}
/* Return 1 if a partial raid LV can be activated redundantly */
static int _partial_raid_lv_is_redundant(const struct logical_volume *lv)
{
struct lv_segment *raid_seg = first_seg(lv);
uint32_t copies;
uint32_t i, s, rebuilds_per_group = 0;
uint32_t failed_components = 0;
if (seg_is_raid10(raid_seg)) {
/* FIXME: We only support 2-way mirrors in RAID10 currently */
copies = 2;
for (i = 0; i < raid_seg->area_count * copies; i++) {
s = i % raid_seg->area_count;
if (!(i % copies))
rebuilds_per_group = 0;
if ((seg_lv(raid_seg, s)->status & PARTIAL_LV) ||
(seg_metalv(raid_seg, s)->status & PARTIAL_LV) ||
lv_is_virtual(seg_lv(raid_seg, s)) ||
lv_is_virtual(seg_metalv(raid_seg, s)))
rebuilds_per_group++;
if (rebuilds_per_group >= copies) {
log_verbose("An entire mirror group has failed in %s.",
display_lvname(lv));
return 0; /* Insufficient redundancy to activate */
}
}
return 1; /* Redundant */
}
for (s = 0; s < raid_seg->area_count; s++) {
if ((seg_lv(raid_seg, s)->status & PARTIAL_LV) ||
(seg_metalv(raid_seg, s)->status & PARTIAL_LV) ||
lv_is_virtual(seg_lv(raid_seg, s)) ||
lv_is_virtual(seg_metalv(raid_seg, s)))
failed_components++;
}
if (failed_components == raid_seg->area_count) {
log_verbose("All components of raid LV %s have failed.",
display_lvname(lv));
return 0; /* Insufficient redundancy to activate */
} else if (raid_seg->segtype->parity_devs &&
(failed_components > raid_seg->segtype->parity_devs)) {
log_verbose("More than %u components from %s %s have failed.",
raid_seg->segtype->parity_devs,
lvseg_name(raid_seg),
display_lvname(lv));
return 0; /* Insufficient redundancy to activate */
}
return 1;
}
/* Sets *data to 1 if the LV cannot be activated without data loss */
static int _lv_may_be_activated_in_degraded_mode(struct logical_volume *lv, void *data)
{
int *not_capable = (int *)data;
uint32_t s;
struct lv_segment *seg;
if (*not_capable)
return 1; /* No further checks needed */
if (!(lv->status & PARTIAL_LV))
return 1;
if (lv_is_raid(lv)) {
*not_capable = !_partial_raid_lv_is_redundant(lv);
return 1;
}
/* Ignore RAID sub-LVs. */
if (lv_is_raid_type(lv))
return 1;
dm_list_iterate_items(seg, &lv->segments)
for (s = 0; s < seg->area_count; s++)
if (seg_type(seg, s) != AREA_LV) {
log_verbose("%s contains a segment incapable of degraded activation",
display_lvname(lv));
*not_capable = 1;
}
return 1;
}
int partial_raid_lv_supports_degraded_activation(const struct logical_volume *clv)
{
int not_capable = 0;
struct logical_volume * lv = (struct logical_volume *)clv; /* drop const */
if (!_lv_may_be_activated_in_degraded_mode(lv, &not_capable) || not_capable)
return_0;
if (!for_each_sub_lv(lv, _lv_may_be_activated_in_degraded_mode, &not_capable)) {
log_error(INTERNAL_ERROR "for_each_sub_lv failure.");
return 0;
}
return !not_capable;
}