2001-10-30 17:32:48 +03:00
/*
2008-01-30 17:00:02 +03:00
* Copyright ( C ) 2001 - 2004 Sistina Software , Inc . All rights reserved .
2014-11-02 19:48:21 +03:00
* Copyright ( C ) 2004 - 2014 Red Hat , Inc . All rights reserved .
2001-10-30 17:32:48 +03:00
*
2004-03-30 23:35:44 +04:00
* This file is part of LVM2 .
2001-10-30 17:32:48 +03:00
*
2004-03-30 23:35:44 +04:00
* This copyrighted material is made available to anyone wishing to use ,
* modify , copy , or redistribute it subject to the terms and conditions
2007-08-21 00:55:30 +04:00
* of the GNU Lesser General Public License v .2 .1 .
2001-10-30 17:32:48 +03:00
*
2007-08-21 00:55:30 +04:00
* You should have received a copy of the GNU Lesser General Public License
2004-03-30 23:35:44 +04:00
* along with this program ; if not , write to the Free Software Foundation ,
* Inc . , 59 Temple Place , Suite 330 , Boston , MA 02111 - 1307 USA
2001-10-30 17:32:48 +03:00
*/
# include "tools.h"
2015-07-06 19:30:18 +03:00
2014-09-19 17:16:26 +04:00
# include "memlock.h"
2001-10-30 17:32:48 +03:00
2015-02-26 21:38:26 +03:00
static int _lvchange_permission ( struct cmd_context * cmd ,
struct logical_volume * lv )
2001-10-30 17:32:48 +03:00
{
2002-12-20 02:25:55 +03:00
uint32_t lv_access ;
2007-01-10 22:56:39 +03:00
struct lvinfo info ;
2015-02-26 21:38:26 +03:00
unsigned info_obtained = 0 ;
2001-10-30 17:32:48 +03:00
2002-12-20 02:25:55 +03:00
lv_access = arg_uint_value ( cmd , permission_ARG , 0 ) ;
2001-10-30 17:32:48 +03:00
2015-02-26 21:38:26 +03:00
if ( lv_is_external_origin ( lv ) ) {
log_error ( " Cannot change permissions of external origin "
" \" %s \" . " , lv - > name ) ;
2001-10-30 17:32:48 +03:00
return 0 ;
}
if ( ! ( lv_access & LVM_WRITE ) & & ! ( lv - > status & LVM_WRITE ) ) {
2015-02-27 16:38:26 +03:00
/* Refresh if it's read-only in metadata but read-write in kernel */
if ( lv_info ( cmd , lv , 0 , & info , 0 , 0 ) & &
( info_obtained = 1 , info . exists ) & & ! info . read_only ) {
log_print_unless_silent ( " Logical volume \" %s \" is already read-only. Refreshing kernel state. " ,
lv - > name ) ;
return lv_refresh ( cmd , lv ) ;
}
2002-02-12 00:00:35 +03:00
log_error ( " Logical volume \" %s \" is already read only " ,
lv - > name ) ;
2001-10-30 17:32:48 +03:00
return 0 ;
}
2015-02-26 21:38:26 +03:00
if ( ( lv_access & LVM_WRITE ) & & ( lv - > status & LVM_WRITE ) ) {
/* Refresh if it's read-write in metadata but read-only in kernel */
if ( lv_info ( cmd , lv , 0 , & info , 0 , 0 ) & &
( info_obtained = 1 , info . exists ) & & info . read_only ) {
log_print_unless_silent ( " Logical volume \" %s \" is already writable. Refreshing kernel state. " ,
lv - > name ) ;
return lv_refresh ( cmd , lv ) ;
}
log_error ( " Logical volume \" %s \" is already writable " ,
lv - > name ) ;
2013-02-05 17:03:43 +04:00
return 0 ;
}
2014-09-16 00:33:53 +04:00
if ( lv_is_mirrored ( lv ) & & vg_is_clustered ( lv - > vg ) & &
2015-02-26 21:38:26 +03:00
( info_obtained | | lv_info ( cmd , lv , 0 , & info , 0 , 0 ) ) & & info . exists ) {
2007-01-10 22:56:39 +03:00
log_error ( " Cannot change permissions of mirror \" %s \" "
" while active. " , lv - > name ) ;
return 0 ;
}
2011-12-01 04:04:21 +04:00
/* Not allowed to change permissions on RAID sub-LVs directly */
2014-09-16 00:33:53 +04:00
if ( lv_is_raid_metadata ( lv ) | | lv_is_raid_image ( lv ) ) {
2011-12-01 04:04:21 +04:00
log_error ( " Cannot change permissions of RAID %s \" %s \" " ,
2014-09-16 00:33:53 +04:00
lv_is_raid_image ( lv ) ? " image " :
2011-12-01 04:04:21 +04:00
" metadata area " , lv - > name ) ;
return 0 ;
}
2012-10-15 15:41:38 +04:00
if ( ! ( lv_access & LVM_WRITE ) & & lv_is_thin_pool ( lv ) ) {
log_error ( " Change permissions of thin pool \" %s \" not "
2012-10-19 15:57:22 +04:00
" yet supported. " , lv - > name ) ;
2012-10-15 15:41:38 +04:00
return 0 ;
}
2001-10-30 17:32:48 +03:00
if ( lv_access & LVM_WRITE ) {
lv - > status | = LVM_WRITE ;
2002-02-12 00:00:35 +03:00
log_verbose ( " Setting logical volume \" %s \" read/write " ,
lv - > name ) ;
2001-10-30 17:32:48 +03:00
} else {
lv - > status & = ~ LVM_WRITE ;
2002-02-12 00:00:35 +03:00
log_verbose ( " Setting logical volume \" %s \" read-only " ,
lv - > name ) ;
2001-10-30 17:32:48 +03:00
}
2014-09-09 16:40:32 +04:00
if ( ! lv_update_and_reload ( lv ) )
2008-01-30 16:19:47 +03:00
return_0 ;
2003-07-05 02:34:56 +04:00
2014-09-09 16:40:32 +04:00
return 1 ;
2001-10-30 17:32:48 +03:00
}
2015-02-26 21:38:26 +03:00
static int _lvchange_pool_update ( struct cmd_context * cmd ,
struct logical_volume * lv )
2012-06-28 16:52:23 +04:00
{
int update = 0 ;
unsigned val ;
2012-08-08 00:24:41 +04:00
thin_discards_t discards ;
2012-06-28 16:52:23 +04:00
if ( ! lv_is_thin_pool ( lv ) ) {
2012-08-07 23:08:14 +04:00
log_error ( " Logical volume \" %s \" is not a thin pool. " , lv - > name ) ;
2012-06-28 16:52:23 +04:00
return 0 ;
}
2012-08-08 00:24:41 +04:00
if ( arg_count ( cmd , discards_ARG ) ) {
2012-08-21 17:51:54 +04:00
discards = ( thin_discards_t ) arg_uint_value ( cmd , discards_ARG , THIN_DISCARDS_IGNORE ) ;
2012-08-08 00:24:41 +04:00
if ( discards ! = first_seg ( lv ) - > discards ) {
2012-12-02 19:30:45 +04:00
if ( ( ( discards = = THIN_DISCARDS_IGNORE ) | |
2012-08-08 00:24:41 +04:00
( first_seg ( lv ) - > discards = = THIN_DISCARDS_IGNORE ) ) & &
2013-02-05 19:49:09 +04:00
pool_is_active ( lv ) )
2013-10-08 02:50:09 +04:00
log_error ( " Cannot change support for discards while pool volume \" %s \" is active. " , lv - > name ) ;
2012-06-28 16:52:23 +04:00
else {
2012-08-08 00:24:41 +04:00
first_seg ( lv ) - > discards = discards ;
2012-06-28 16:52:23 +04:00
update + + ;
}
} else
2012-08-08 00:24:41 +04:00
log_error ( " Logical volume \" %s \" already uses --discards %s. " ,
lv - > name , get_pool_discards_name ( discards ) ) ;
2012-06-28 16:52:23 +04:00
}
if ( arg_count ( cmd , zero_ARG ) ) {
val = arg_uint_value ( cmd , zero_ARG , 1 ) ;
if ( val ! = first_seg ( lv ) - > zero_new_blocks ) {
first_seg ( lv ) - > zero_new_blocks = val ;
update + + ;
} else
log_error ( " Logical volume \" %s \" already %szero new blocks. " ,
lv - > name , val ? " " : " does not " ) ;
}
if ( ! update )
return 0 ;
2014-09-09 16:40:32 +04:00
if ( ! lv_update_and_reload_origin ( lv ) )
2012-06-28 16:52:23 +04:00
return_0 ;
2014-09-09 16:40:32 +04:00
return 1 ;
2012-06-28 16:52:23 +04:00
}
2015-02-26 21:38:26 +03:00
static int _lvchange_monitoring ( struct cmd_context * cmd ,
struct logical_volume * lv )
2006-05-12 23:16:48 +04:00
{
struct lvinfo info ;
2012-03-23 13:58:04 +04:00
if ( ! lv_info ( cmd , lv , lv_is_thin_pool ( lv ) ? 1 : 0 ,
& info , 0 , 0 ) | | ! info . exists ) {
2006-05-12 23:16:48 +04:00
log_error ( " Logical volume, %s, is not active " , lv - > name ) ;
return 0 ;
}
2007-01-20 01:21:45 +03:00
/* do not monitor pvmove lv's */
2014-09-16 00:33:53 +04:00
if ( lv_is_pvmove ( lv ) )
2006-05-12 23:16:48 +04:00
return 1 ;
2007-01-25 02:43:27 +03:00
if ( ( dmeventd_monitor_mode ( ) ! = DMEVENTD_MONITOR_IGNORE ) & &
2010-08-17 20:25:32 +04:00
! monitor_dev_for_events ( cmd , lv , 0 , dmeventd_monitor_mode ( ) ) )
2010-08-17 02:54:35 +04:00
return_0 ;
2006-05-12 23:16:48 +04:00
2007-01-20 01:21:45 +03:00
return 1 ;
2006-05-12 23:16:48 +04:00
}
2015-02-26 21:38:26 +03:00
static int _lvchange_background_polling ( struct cmd_context * cmd ,
struct logical_volume * lv )
2010-01-05 23:56:51 +03:00
{
struct lvinfo info ;
2010-08-17 20:25:32 +04:00
if ( ! lv_info ( cmd , lv , 0 , & info , 0 , 0 ) | | ! info . exists ) {
2010-01-05 23:56:51 +03:00
log_error ( " Logical volume, %s, is not active " , lv - > name ) ;
return 0 ;
}
if ( background_polling ( ) )
lv_spawn_background_polling ( cmd , lv ) ;
return 1 ;
}
2012-06-27 15:48:31 +04:00
static int _lvchange_activate ( struct cmd_context * cmd , struct logical_volume * lv )
2001-10-30 17:32:48 +03:00
{
2013-04-21 14:49:58 +04:00
activation_change_t activate ;
2001-10-30 17:32:48 +03:00
2013-04-21 14:49:58 +04:00
activate = ( activation_change_t ) arg_uint_value ( cmd , activate_ARG , CHANGE_AY ) ;
2001-10-30 17:32:48 +03:00
2015-02-25 20:33:11 +03:00
/*
* We can get here in the odd case where an LV is already active in
* a foreign VG , which allows the VG to be accessed by lvchange - a
* so the LV can be deactivated .
*/
2015-03-04 01:45:16 +03:00
if ( lv - > vg - > system_id & & lv - > vg - > system_id [ 0 ] & &
cmd - > system_id & & cmd - > system_id [ 0 ] & &
2015-02-25 20:33:11 +03:00
strcmp ( lv - > vg - > system_id , cmd - > system_id ) & &
is_change_activating ( activate ) ) {
log_error ( " Cannot activate LVs in a foreign VG. " ) ;
return ECMD_FAILED ;
}
2014-02-19 14:02:16 +04:00
if ( lv_activation_skip ( lv , activate , arg_count ( cmd , ignoreactivationskip_ARG ) ) )
2013-07-11 14:44:36 +04:00
return 1 ;
2011-11-18 23:22:49 +04:00
if ( lv_is_cow ( lv ) & & ! lv_is_virtual_origin ( origin_from_cow ( lv ) ) )
lv = origin_from_cow ( lv ) ;
2013-04-11 15:51:08 +04:00
if ( ( activate = = CHANGE_AAY ) & &
! lv_passes_auto_activation_filter ( cmd , lv ) )
return 1 ;
2012-06-27 18:43:20 +04:00
2013-04-11 15:51:08 +04:00
if ( ! lv_change_activate ( cmd , lv , activate ) )
return_0 ;
2002-01-08 01:36:12 +03:00
2015-04-08 13:05:14 +03:00
/*
* FIXME : lvchange should defer background polling in a similar
* way as vgchange does . First activate all relevant LVs
* initate background polling later ( for all actually
* activated LVs ) . So we can avoid duplicate background
* polling for pvmove ( 2 or more locked LVs on single pvmove
* LV )
*/
if ( background_polling ( ) & & is_change_activating ( activate ) & &
( lv_is_pvmove ( lv ) | | lv_is_locked ( lv ) | | lv_is_converting ( lv ) | |
lv_is_merging ( lv ) ) )
lv_spawn_background_polling ( cmd , lv ) ;
2001-10-30 17:32:48 +03:00
return 1 ;
}
2012-09-11 22:01:05 +04:00
static int detach_metadata_devices ( struct lv_segment * seg , struct dm_list * list )
{
2012-09-11 22:09:35 +04:00
uint32_t s ;
uint32_t num_meta_lvs ;
2012-09-11 22:01:05 +04:00
struct lv_list * lvl ;
2012-09-11 22:09:35 +04:00
num_meta_lvs = seg_is_raid ( seg ) ? seg - > area_count : ! ! seg - > log_lv ;
if ( ! num_meta_lvs )
return_0 ;
2012-09-11 22:01:05 +04:00
2013-04-25 12:15:13 +04:00
if ( ! ( lvl = dm_pool_alloc ( seg - > lv - > vg - > vgmem , sizeof ( * lvl ) * num_meta_lvs ) ) )
2012-09-11 22:01:05 +04:00
return_0 ;
2012-09-11 22:09:35 +04:00
if ( seg_is_raid ( seg ) ) {
for ( s = 0 ; s < seg - > area_count ; s + + ) {
if ( ! seg_metalv ( seg , s ) )
return_0 ; /* Trap this future possibility */
lvl [ s ] . lv = seg_metalv ( seg , s ) ;
lv_set_visible ( lvl [ s ] . lv ) ;
dm_list_add ( list , & lvl [ s ] . list ) ;
}
return 1 ;
}
lvl [ 0 ] . lv = detach_mirror_log ( seg ) ;
dm_list_add ( list , & lvl [ 0 ] . list ) ;
2012-09-11 22:01:05 +04:00
return 1 ;
}
static int attach_metadata_devices ( struct lv_segment * seg , struct dm_list * list )
{
2013-04-23 16:06:40 +04:00
struct lv_list * lvl ;
2012-09-11 22:01:05 +04:00
if ( seg_is_raid ( seg ) ) {
2013-04-23 16:06:40 +04:00
dm_list_iterate_items ( lvl , list )
2012-09-11 22:09:35 +04:00
lv_set_hidden ( lvl - > lv ) ;
return 1 ;
2012-09-11 22:01:05 +04:00
}
dm_list_iterate_items ( lvl , list )
break ; /* get first item */
2013-04-23 16:06:40 +04:00
if ( ! attach_mirror_log ( seg , lvl - > lv ) )
2012-09-11 22:01:05 +04:00
return_0 ;
return 1 ;
}
2013-04-12 00:33:59 +04:00
/*
* lvchange_refresh
* @ cmd
* @ lv
*
* Suspend and resume a logical volume .
*/
2015-02-26 21:38:26 +03:00
static int _lvchange_refresh ( struct cmd_context * cmd , struct logical_volume * lv )
2013-04-12 00:33:59 +04:00
{
log_verbose ( " Refreshing logical volume \" %s \" (if active) " , lv - > name ) ;
return lv_refresh ( cmd , lv ) ;
}
2013-09-11 01:33:22 +04:00
static int _reactivate_lv ( struct logical_volume * lv ,
int active , int exclusive )
{
struct cmd_context * cmd = lv - > vg - > cmd ;
if ( ! active )
return 1 ;
if ( exclusive )
return activate_lv_excl_local ( cmd , lv ) ;
return activate_lv ( cmd , lv ) ;
}
2013-04-12 00:33:59 +04:00
/*
* lvchange_resync
* @ cmd
* @ lv
*
* Force a mirror or RAID array to undergo a complete initializing resync .
*/
2015-02-26 21:38:26 +03:00
static int _lvchange_resync ( struct cmd_context * cmd , struct logical_volume * lv )
2006-10-24 03:03:55 +04:00
{
int active = 0 ;
2013-09-11 01:33:22 +04:00
int exclusive = 0 ;
2008-05-21 18:10:11 +04:00
int monitored ;
2012-09-11 22:01:05 +04:00
struct lv_segment * seg = first_seg ( lv ) ;
struct dm_list device_list ;
struct lv_list * lvl ;
dm_list_init ( & device_list ) ;
2006-10-24 03:03:55 +04:00
2014-09-16 03:13:46 +04:00
if ( ! seg_is_mirror ( seg ) & & ! seg_is_raid ( seg ) ) {
2012-09-11 22:09:35 +04:00
log_error ( " Unable to resync %s. It is not RAID or mirrored. " ,
2006-10-24 03:03:55 +04:00
lv - > name ) ;
2012-10-15 15:42:18 +04:00
return 0 ;
2006-10-24 03:03:55 +04:00
}
2014-09-16 00:33:53 +04:00
if ( lv_is_pvmove ( lv ) ) {
2006-10-24 03:03:55 +04:00
log_error ( " Unable to resync pvmove volume %s " , lv - > name ) ;
return 0 ;
}
2014-09-16 00:33:53 +04:00
if ( lv_is_locked ( lv ) ) {
2006-10-24 03:03:55 +04:00
log_error ( " Unable to resync locked volume %s " , lv - > name ) ;
return 0 ;
}
2014-09-24 12:05:26 +04:00
if ( lv_is_active_locally ( lv ) ) {
if ( ! lv_check_not_in_use ( lv ) ) {
2006-10-24 03:03:55 +04:00
log_error ( " Can't resync open logical volume \" %s \" " ,
lv - > name ) ;
2009-05-20 13:55:33 +04:00
return 0 ;
2006-10-24 03:03:55 +04:00
}
2014-09-24 18:27:35 +04:00
if ( ! arg_count ( cmd , yes_ARG ) & &
yes_no_prompt ( " Do you really want to deactivate "
" logical volume %s to resync it? [y/n]: " ,
lv - > name ) = = ' n ' ) {
log_error ( " Logical volume \" %s \" not resynced " ,
lv - > name ) ;
return 0 ;
2006-10-24 19:30:33 +04:00
}
2014-09-24 18:27:35 +04:00
active = 1 ;
if ( lv_is_active_exclusive_locally ( lv ) )
exclusive = 1 ;
2006-10-24 03:03:55 +04:00
}
2013-09-11 01:33:22 +04:00
if ( seg_is_raid ( seg ) & & active & & ! exclusive ) {
log_error ( " RAID logical volume %s/%s cannot be active remotely. " ,
lv - > vg - > name , lv - > name ) ;
return 0 ;
}
2008-05-21 18:10:11 +04:00
/* Activate exclusively to ensure no nodes still have LV active */
monitored = dmeventd_monitor_mode ( ) ;
2012-10-09 14:20:47 +04:00
if ( monitored ! = DMEVENTD_MONITOR_IGNORE )
init_dmeventd_monitor ( 0 ) ;
2008-05-21 18:10:11 +04:00
2006-10-24 03:03:55 +04:00
if ( ! deactivate_lv ( cmd , lv ) ) {
log_error ( " Unable to deactivate %s for resync " , lv - > name ) ;
return 0 ;
}
2009-05-20 13:55:33 +04:00
if ( vg_is_clustered ( lv - > vg ) & & lv_is_active ( lv ) ) {
log_error ( " Can't get exclusive access to clustered volume %s " ,
lv - > name ) ;
return 0 ;
}
2012-10-09 14:20:47 +04:00
if ( monitored ! = DMEVENTD_MONITOR_IGNORE )
init_dmeventd_monitor ( monitored ) ;
2012-09-11 22:09:35 +04:00
init_mirror_in_sync ( 0 ) ;
2008-05-21 18:10:11 +04:00
2012-09-11 22:09:35 +04:00
log_very_verbose ( " Starting resync of %s%s%s%s \" %s \" " ,
2006-10-24 03:03:55 +04:00
( active ) ? " active " : " " ,
2008-04-10 21:09:32 +04:00
vg_is_clustered ( lv - > vg ) ? " clustered " : " " ,
2012-09-11 22:09:35 +04:00
( seg - > log_lv ) ? " disk-logged " :
seg_is_raid ( seg ) ? " " : " core-logged " ,
2014-10-20 20:40:39 +04:00
lvseg_name ( seg ) , lv - > name ) ;
2006-10-24 03:03:55 +04:00
/*
2012-09-11 22:01:05 +04:00
* If this mirror has a core log ( i . e . ! seg - > log_lv ) ,
2006-10-24 03:03:55 +04:00
* then simply deactivating / activating will cause
* it to reset the sync status . We only need to
* worry about persistent logs .
*/
2012-09-11 22:09:35 +04:00
if ( ! seg_is_raid ( seg ) & & ! seg - > log_lv ) {
2012-09-19 20:09:32 +04:00
if ( lv - > status & LV_NOTSYNCED ) {
2012-09-11 21:55:17 +04:00
lv - > status & = ~ LV_NOTSYNCED ;
log_very_verbose ( " Updating logical volume \" %s \" "
" on disk(s) " , lv - > name ) ;
if ( ! vg_write ( lv - > vg ) | | ! vg_commit ( lv - > vg ) ) {
log_error ( " Failed to update metadata on disk. " ) ;
return 0 ;
}
}
2013-09-11 01:33:22 +04:00
if ( ! _reactivate_lv ( lv , active , exclusive ) ) {
2006-10-24 03:03:55 +04:00
log_error ( " Failed to reactivate %s to resynchronize "
" mirror " , lv - > name ) ;
return 0 ;
}
2012-09-11 21:55:17 +04:00
2006-10-24 03:03:55 +04:00
return 1 ;
}
2012-09-11 21:55:17 +04:00
/*
* Now we handle mirrors with log devices
*/
2011-03-29 16:51:57 +04:00
lv - > status & = ~ LV_NOTSYNCED ;
2006-10-24 03:03:55 +04:00
2012-09-11 22:01:05 +04:00
/* Separate mirror log or metadata devices so we can clear them */
if ( ! detach_metadata_devices ( seg , & device_list ) ) {
log_error ( " Failed to clear %s %s for %s " ,
seg - > segtype - > name , seg_is_raid ( seg ) ?
" metadata area " : " mirror log " , lv - > name ) ;
return 0 ;
}
2006-10-24 03:03:55 +04:00
2014-09-19 17:10:21 +04:00
if ( ! vg_write ( lv - > vg ) | | ! vg_commit ( lv - > vg ) ) {
log_error ( " Failed to update intermediate VG metadata on disk. " ) ;
2013-09-11 01:33:22 +04:00
if ( ! _reactivate_lv ( lv , active , exclusive ) )
2012-09-11 21:55:17 +04:00
stack ;
return 0 ;
}
2009-04-21 18:31:57 +04:00
2014-09-19 17:16:26 +04:00
/* No backup for intermediate metadata, so just unlock memory */
memlock_unlock ( lv - > vg - > cmd ) ;
2006-10-24 03:03:55 +04:00
2012-09-11 22:01:05 +04:00
dm_list_iterate_items ( lvl , & device_list ) {
2013-09-11 01:33:22 +04:00
if ( ! activate_lv_excl_local ( cmd , lvl - > lv ) ) {
log_error ( " Unable to activate %s for %s clearing " ,
lvl - > lv - > name , ( seg_is_raid ( seg ) ) ?
" metadata area " : " mirror log " ) ;
2012-09-11 22:01:05 +04:00
return 0 ;
}
2006-10-24 03:03:55 +04:00
2013-11-28 14:22:24 +04:00
if ( ! wipe_lv ( lvl - > lv , ( struct wipe_params )
{ . do_zero = 1 , . zero_sectors = lvl - > lv - > size } ) ) {
2012-09-11 22:01:05 +04:00
log_error ( " Unable to reset sync status for %s " ,
lv - > name ) ;
if ( ! deactivate_lv ( cmd , lvl - > lv ) )
log_error ( " Failed to deactivate log LV after "
" wiping failed " ) ;
return 0 ;
}
if ( ! deactivate_lv ( cmd , lvl - > lv ) ) {
log_error ( " Unable to deactivate %s LV %s "
" after wiping for resync " ,
( seg_is_raid ( seg ) ) ? " metadata " : " log " ,
lvl - > lv - > name ) ;
return 0 ;
}
2012-09-11 21:55:17 +04:00
}
2006-10-24 03:03:55 +04:00
2015-06-30 20:54:38 +03:00
/* Wait until devices are away */
if ( ! sync_local_dev_names ( lv - > vg - > cmd ) ) {
log_error ( " Failed to sync local devices after updating %s " ,
display_lvname ( lv ) ) ;
return 0 ;
}
2014-09-19 17:16:26 +04:00
2012-09-11 22:01:05 +04:00
/* Put metadata sub-LVs back in place */
if ( ! attach_metadata_devices ( seg , & device_list ) ) {
log_error ( " Failed to reattach %s device after clearing " ,
( seg_is_raid ( seg ) ) ? " metadata " : " log " ) ;
2012-09-11 21:55:17 +04:00
return 0 ;
2006-10-24 03:03:55 +04:00
}
if ( ! vg_write ( lv - > vg ) | | ! vg_commit ( lv - > vg ) ) {
log_error ( " Failed to update metadata on disk. " ) ;
return 0 ;
}
2013-09-11 01:33:22 +04:00
if ( ! _reactivate_lv ( lv , active , exclusive ) ) {
2014-09-19 17:16:26 +04:00
backup ( lv - > vg ) ;
2013-09-11 01:33:22 +04:00
log_error ( " Failed to reactivate %s after resync " ,
lv - > name ) ;
2006-10-24 03:03:55 +04:00
return 0 ;
}
2014-09-19 17:16:26 +04:00
backup ( lv - > vg ) ;
2006-10-24 03:03:55 +04:00
return 1 ;
}
2015-02-26 21:38:26 +03:00
static int _lvchange_alloc ( struct cmd_context * cmd , struct logical_volume * lv )
2001-10-30 17:32:48 +03:00
{
2014-10-11 20:17:46 +04:00
int want_contiguous = arg_int_value ( cmd , contiguous_ARG , 0 ) ;
alloc_policy_t alloc = ( alloc_policy_t )
arg_uint_value ( cmd , alloc_ARG , ( want_contiguous )
? ALLOC_CONTIGUOUS : ALLOC_INHERIT ) ;
2001-10-30 17:32:48 +03:00
2004-05-19 02:12:53 +04:00
if ( alloc = = lv - > alloc ) {
2002-01-30 18:04:48 +03:00
log_error ( " Allocation policy of logical volume \" %s \" is "
2004-05-19 02:12:53 +04:00
" already %s " , lv - > name , get_alloc_string ( alloc ) ) ;
2001-10-30 17:32:48 +03:00
return 0 ;
}
2004-05-19 02:12:53 +04:00
lv - > alloc = alloc ;
2001-10-30 17:32:48 +03:00
2004-05-19 02:12:53 +04:00
/* FIXME If contiguous, check existing extents already are */
2001-10-30 17:32:48 +03:00
2004-05-19 02:12:53 +04:00
log_verbose ( " Setting contiguous allocation policy for \" %s \" to %s " ,
lv - > name , get_alloc_string ( alloc ) ) ;
2001-10-30 17:32:48 +03:00
2003-07-05 02:34:56 +04:00
log_very_verbose ( " Updating logical volume \" %s \" on disk(s) " , lv - > name ) ;
2004-05-19 02:12:53 +04:00
2009-04-21 18:31:57 +04:00
/* No need to suspend LV for this change */
if ( ! vg_write ( lv - > vg ) | | ! vg_commit ( lv - > vg ) )
2008-01-30 16:19:47 +03:00
return_0 ;
2003-07-05 02:34:56 +04:00
backup ( lv - > vg ) ;
2001-10-30 17:32:48 +03:00
return 1 ;
}
2015-02-26 21:38:26 +03:00
static int _lvchange_errorwhenfull ( struct cmd_context * cmd ,
struct logical_volume * lv )
2015-01-15 17:20:08 +03:00
{
unsigned ewf = arg_int_value ( cmd , errorwhenfull_ARG , 0 ) ;
if ( ewf = = lv_is_error_when_full ( lv ) ) {
log_error ( " Error when full is already %sset for %s. " ,
( ewf ) ? " " : " un " , display_lvname ( lv ) ) ;
return 0 ;
}
if ( ewf )
lv - > status | = LV_ERROR_WHEN_FULL ;
else
lv - > status & = ~ LV_ERROR_WHEN_FULL ;
if ( ! lv_update_and_reload ( lv ) )
return_0 ;
return 1 ;
}
2015-02-26 21:38:26 +03:00
static int _lvchange_readahead ( struct cmd_context * cmd ,
struct logical_volume * lv )
2001-10-30 17:32:48 +03:00
{
2006-05-10 01:23:51 +04:00
unsigned read_ahead = 0 ;
2007-12-05 22:24:32 +03:00
unsigned pagesize = ( unsigned ) lvm_getpagesize ( ) > > SECTOR_SHIFT ;
2001-10-30 17:32:48 +03:00
2002-12-20 02:25:55 +03:00
read_ahead = arg_uint_value ( cmd , readahead_ARG , 0 ) ;
2001-10-30 17:32:48 +03:00
2007-11-09 19:51:54 +03:00
if ( read_ahead ! = DM_READ_AHEAD_AUTO & &
( lv - > vg - > fid - > fmt - > features & FMT_RESTRICTED_READAHEAD ) & &
( read_ahead < 2 | | read_ahead > 120 ) ) {
log_error ( " Metadata only supports readahead values between 2 and 120. " ) ;
2001-10-30 17:32:48 +03:00
return 0 ;
}
2007-12-05 22:24:32 +03:00
if ( read_ahead ! = DM_READ_AHEAD_AUTO & &
read_ahead ! = DM_READ_AHEAD_NONE & & read_ahead % pagesize ) {
2009-06-07 02:00:20 +04:00
if ( read_ahead < pagesize )
read_ahead = pagesize ;
else
read_ahead = ( read_ahead / pagesize ) * pagesize ;
log_warn ( " WARNING: Overriding readahead to %u sectors, a multiple "
" of %uK page size. " , read_ahead , pagesize > > 1 ) ;
2007-12-05 22:24:32 +03:00
}
2001-10-30 17:32:48 +03:00
if ( lv - > read_ahead = = read_ahead ) {
2008-06-18 15:32:14 +04:00
if ( read_ahead = = DM_READ_AHEAD_AUTO )
log_error ( " Read ahead is already auto for \" %s \" " , lv - > name ) ;
else
log_error ( " Read ahead is already %u for \" %s \" " ,
read_ahead , lv - > name ) ;
2001-10-30 17:32:48 +03:00
return 0 ;
}
lv - > read_ahead = read_ahead ;
2002-02-21 00:30:27 +03:00
2002-02-12 00:00:35 +03:00
log_verbose ( " Setting read ahead to %u for \" %s \" " , read_ahead ,
lv - > name ) ;
2001-10-30 17:32:48 +03:00
2014-09-09 16:40:32 +04:00
if ( ! lv_update_and_reload ( lv ) )
2008-01-30 16:19:47 +03:00
return_0 ;
2003-07-05 02:34:56 +04:00
2014-09-09 16:40:32 +04:00
return 1 ;
2001-10-30 17:32:48 +03:00
}
2002-02-01 20:54:39 +03:00
2015-02-26 21:38:26 +03:00
static int _lvchange_persistent ( struct cmd_context * cmd ,
struct logical_volume * lv )
2002-02-01 20:54:39 +03:00
{
2014-09-19 16:57:02 +04:00
enum activation_change activate = CHANGE_AN ;
2015-03-05 23:00:44 +03:00
/* The LV lock in lvmlockd should remain as it is. */
cmd - > lockd_lv_disable = 1 ;
2014-09-27 20:53:08 +04:00
if ( ! get_and_validate_major_minor ( cmd , lv - > vg - > fid - > fmt ,
& lv - > major , & lv - > minor ) )
2014-09-19 16:57:02 +04:00
return_0 ;
2002-02-25 15:56:16 +03:00
2014-09-19 16:57:02 +04:00
if ( lv - > minor = = - 1 ) {
2002-02-01 20:54:39 +03:00
if ( ! ( lv - > status & FIXED_MINOR ) ) {
2014-09-19 16:57:02 +04:00
log_error ( " Minor number is already not persistent for %s. " ,
display_lvname ( lv ) ) ;
2002-02-01 20:54:39 +03:00
return 0 ;
}
lv - > status & = ~ FIXED_MINOR ;
2014-09-19 16:57:02 +04:00
log_verbose ( " Disabling persistent device number for %s. " ,
display_lvname ( lv ) ) ;
2002-02-01 20:54:39 +03:00
} else {
2014-09-19 16:57:02 +04:00
if ( lv_is_active ( lv ) ) {
if ( ! arg_count ( cmd , force_ARG ) & &
! arg_count ( cmd , yes_ARG ) & &
yes_no_prompt ( " Logical volume %s will be "
" deactivated temporarily. "
" Continue? [y/n]: " , lv - > name ) = = ' n ' ) {
log_error ( " %s device number not changed. " ,
display_lvname ( lv ) ) ;
return 0 ;
}
2012-08-06 20:01:01 +04:00
2014-09-19 16:57:02 +04:00
activate = CHANGE_AEY ;
if ( vg_is_clustered ( lv - > vg ) & &
locking_is_clustered ( ) & &
locking_supports_remote_queries ( ) & &
! lv_is_active_exclusive_locally ( lv ) ) {
/* Reliable reactivate only locally */
log_print_unless_silent ( " Remotely active LV %s needs "
" individual reactivation. " ,
display_lvname ( lv ) ) ;
activate = CHANGE_ALY ;
}
2003-07-11 21:10:19 +04:00
}
2007-06-15 14:11:14 +04:00
2014-09-19 16:57:02 +04:00
/* Ensuring LV is not active */
2005-08-15 16:00:04 +04:00
if ( ! deactivate_lv ( cmd , lv ) ) {
2014-09-19 16:57:02 +04:00
log_error ( " Cannot deactivate %s. " , display_lvname ( lv ) ) ;
2002-03-11 22:02:28 +03:00
return 0 ;
}
2002-02-01 20:54:39 +03:00
lv - > status | = FIXED_MINOR ;
2014-09-19 16:57:02 +04:00
log_verbose ( " Setting persistent device number to (%d, %d) for %s. " ,
lv - > major , lv - > minor , display_lvname ( lv ) ) ;
2002-02-01 20:54:39 +03:00
}
2014-09-19 16:57:02 +04:00
log_very_verbose ( " Updating logical volume %s on disk(s). " ,
display_lvname ( lv ) ) ;
2009-04-21 18:31:57 +04:00
if ( ! vg_write ( lv - > vg ) | | ! vg_commit ( lv - > vg ) )
2008-01-30 16:19:47 +03:00
return_0 ;
2003-07-05 02:34:56 +04:00
2014-09-19 16:57:02 +04:00
if ( activate ! = CHANGE_AN ) {
log_verbose ( " Re-activating logical volume %s " , display_lvname ( lv ) ) ;
2014-11-05 17:14:58 +03:00
if ( ! lv_active_change ( cmd , lv , activate , 0 ) ) {
2014-09-19 16:57:02 +04:00
log_error ( " %s: reactivation failed " , display_lvname ( lv ) ) ;
backup ( lv - > vg ) ;
2006-07-10 23:39:14 +04:00
return 0 ;
}
2002-04-24 22:20:51 +04:00
}
2002-02-21 00:30:27 +03:00
2014-09-19 16:57:02 +04:00
backup ( lv - > vg ) ;
2002-02-01 20:54:39 +03:00
return 1 ;
}
2002-11-18 17:04:08 +03:00
2015-02-26 21:38:26 +03:00
static int _lvchange_cachepolicy ( struct cmd_context * cmd , struct logical_volume * lv )
2014-11-19 20:39:58 +03:00
{
2015-07-15 12:06:40 +03:00
const char * name ;
struct dm_config_tree * settings = NULL ;
2014-11-19 20:39:58 +03:00
int r = 0 ;
if ( ! lv_is_cache ( lv ) & & ! lv_is_cache_pool ( lv ) ) {
log_error ( " LV %s is not a cache LV. " , lv - > name ) ;
log_error ( " Only cache or cache pool devices can have --cachepolicy set. " ) ;
2015-07-15 12:06:40 +03:00
goto out ;
2014-11-19 20:39:58 +03:00
}
2015-07-15 12:06:40 +03:00
if ( ! get_cache_policy_params ( cmd , & name , & settings ) )
2014-11-19 20:39:58 +03:00
goto_out ;
2015-07-15 12:06:40 +03:00
if ( ! lv_cache_set_policy ( lv , name , settings ) )
2014-11-19 20:39:58 +03:00
goto_out ;
if ( ! lv_update_and_reload ( lv ) )
goto_out ;
r = 1 ;
out :
2015-07-15 12:06:40 +03:00
if ( settings )
dm_config_destroy ( settings ) ;
2014-11-19 20:39:58 +03:00
return r ;
}
2015-02-26 21:38:26 +03:00
static int _lvchange_tag ( struct cmd_context * cmd , struct logical_volume * lv , int arg )
2004-03-08 20:19:15 +03:00
{
2011-01-24 16:38:31 +03:00
if ( ! change_tag ( cmd , NULL , lv , NULL , arg ) )
return_0 ;
2004-03-08 20:19:15 +03:00
log_very_verbose ( " Updating logical volume \" %s \" on disk(s) " , lv - > name ) ;
/* No need to suspend LV for this change */
2009-04-21 18:31:57 +04:00
if ( ! vg_write ( lv - > vg ) | | ! vg_commit ( lv - > vg ) )
2008-01-30 16:19:47 +03:00
return_0 ;
2004-03-08 20:19:15 +03:00
2009-04-21 18:31:57 +04:00
backup ( lv - > vg ) ;
2004-03-08 20:19:15 +03:00
return 1 ;
}
2005-03-22 01:55:12 +03:00
2015-02-26 21:38:26 +03:00
static int _lvchange_writemostly ( struct logical_volume * lv )
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
{
int s , pv_count , i = 0 ;
char * * pv_names ;
const char * tmp_str ;
2013-07-19 23:37:43 +04:00
size_t tmp_str_len ;
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
struct pv_list * pvl ;
struct arg_value_group_list * group ;
struct cmd_context * cmd = lv - > vg - > cmd ;
struct lv_segment * raid_seg = first_seg ( lv ) ;
2014-09-24 17:24:41 +04:00
if ( strcmp ( raid_seg - > segtype - > name , SEG_TYPE_NAME_RAID1 ) ) {
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
log_error ( " --write%s can only be used with 'raid1' segment type " ,
arg_count ( cmd , writemostly_ARG ) ? " mostly " : " behind " ) ;
return 0 ;
}
if ( arg_count ( cmd , writebehind_ARG ) )
raid_seg - > writebehind = arg_uint_value ( cmd , writebehind_ARG , 0 ) ;
2014-08-21 17:38:18 +04:00
if ( ( pv_count = arg_count ( cmd , writemostly_ARG ) ) ) {
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
/* writemostly can be specified more than once */
2013-04-25 12:15:13 +04:00
pv_names = dm_pool_alloc ( lv - > vg - > vgmem , sizeof ( char * ) * pv_count ) ;
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 ( ! pv_names )
return_0 ;
dm_list_iterate_items ( group , & cmd - > arg_value_groups ) {
if ( ! grouped_arg_is_set ( group - > arg_values ,
writemostly_ARG ) )
continue ;
if ( ! ( tmp_str = grouped_arg_str_value ( group - > arg_values ,
writemostly_ARG ,
NULL ) ) )
return_0 ;
/*
* Writemostly PV specifications can be :
* < PV > - Turn on writemostly
* < PV > : t - Toggle writemostly
* < PV > : n - Turn off writemostly
* < PV > : y - Turn on writemostly
*
* We allocate strlen + 3 to add our own ' : { t | n | y } ' if
* not present plus the trailing ' \0 ' .
*/
2013-07-19 23:37:43 +04:00
tmp_str_len = strlen ( tmp_str ) ;
if ( ! ( pv_names [ i ] = dm_pool_zalloc ( lv - > vg - > vgmem , tmp_str_len + 3 ) ) )
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
return_0 ;
2013-10-24 07:44:04 +04:00
if ( ( tmp_str_len < 3 ) | |
( tmp_str [ tmp_str_len - 2 ] ! = ' : ' ) )
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
/* Default to 'y' if no mode specified */
sprintf ( pv_names [ i ] , " %s:y " , tmp_str ) ;
else
sprintf ( pv_names [ i ] , " %s " , tmp_str ) ;
i + + ;
}
for ( i = 0 ; i < pv_count ; i + + )
pv_names [ i ] [ strlen ( pv_names [ i ] ) - 2 ] = ' \0 ' ;
for ( i = 0 ; i < pv_count ; i + + ) {
if ( ! ( pvl = find_pv_in_vg ( lv - > vg , pv_names [ i ] ) ) ) {
log_error ( " %s not found in volume group, %s " ,
pv_names [ i ] , lv - > vg - > name ) ;
return 0 ;
}
for ( s = 0 ; s < raid_seg - > area_count ; s + + ) {
/*
* We don ' t bother checking the metadata area ,
* since writemostly only affects the data areas .
*/
2014-08-29 13:58:16 +04:00
if ( seg_type ( raid_seg , s ) = = AREA_UNASSIGNED )
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
continue ;
if ( lv_is_on_pv ( seg_lv ( raid_seg , s ) , pvl - > pv ) ) {
if ( pv_names [ i ] [ strlen ( pv_names [ i ] ) + 1 ] = = ' y ' )
seg_lv ( raid_seg , s ) - > status | =
LV_WRITEMOSTLY ;
else if ( pv_names [ i ] [ strlen ( pv_names [ i ] ) + 1 ] = = ' n ' )
seg_lv ( raid_seg , s ) - > status & =
~ LV_WRITEMOSTLY ;
else if ( pv_names [ i ] [ strlen ( pv_names [ i ] ) + 1 ] = = ' t ' )
seg_lv ( raid_seg , s ) - > status ^ =
LV_WRITEMOSTLY ;
else
return_0 ;
}
}
}
}
2014-09-09 16:40:32 +04:00
if ( ! lv_update_and_reload ( lv ) )
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
return_0 ;
return 1 ;
}
2015-02-26 21:38:26 +03:00
static int _lvchange_recovery_rate ( struct logical_volume * lv )
2013-05-31 20:25:52 +04:00
{
struct cmd_context * cmd = lv - > vg - > cmd ;
struct lv_segment * raid_seg = first_seg ( lv ) ;
2013-08-12 21:40:52 +04:00
if ( ! seg_is_raid ( raid_seg ) ) {
2013-05-31 20:25:52 +04:00
log_error ( " Unable to change the recovery rate of non-RAID "
" logical volume. " ) ;
return 0 ;
}
if ( arg_count ( cmd , minrecoveryrate_ARG ) )
raid_seg - > min_recovery_rate =
2013-08-12 21:40:52 +04:00
arg_uint_value ( cmd , minrecoveryrate_ARG , 0 ) / 2 ;
2013-05-31 20:25:52 +04:00
if ( arg_count ( cmd , maxrecoveryrate_ARG ) )
raid_seg - > max_recovery_rate =
2013-08-12 21:40:52 +04:00
arg_uint_value ( cmd , maxrecoveryrate_ARG , 0 ) / 2 ;
2013-05-31 20:25:52 +04:00
if ( raid_seg - > max_recovery_rate & &
( raid_seg - > max_recovery_rate < raid_seg - > min_recovery_rate ) ) {
log_error ( " Minumum recovery rate cannot "
" be higher than maximum. " ) ;
return 0 ;
}
2014-09-09 16:40:32 +04:00
if ( ! lv_update_and_reload ( lv ) )
2013-05-31 20:25:52 +04:00
return_0 ;
return 1 ;
}
2015-02-26 21:38:26 +03:00
static int _lvchange_profile ( struct logical_volume * lv )
2013-06-25 14:33:06 +04:00
{
const char * old_profile_name , * new_profile_name ;
struct profile * new_profile ;
old_profile_name = lv - > profile ? lv - > profile - > name : " (inherited) " ;
if ( arg_count ( lv - > vg - > cmd , detachprofile_ARG ) ) {
new_profile_name = " (inherited) " ;
lv - > profile = NULL ;
} else {
config: differentiate command and metadata profiles and consolidate profile handling code
- When defining configuration source, the code now uses separate
CONFIG_PROFILE_COMMAND and CONFIG_PROFILE_METADATA markers
(before, it was just CONFIG_PROFILE that did not make the
difference between the two). This helps when checking the
configuration if it contains correct set of options which
are all in either command-profilable or metadata-profilable
group without mixing these groups together - so it's a firm
distinction. The "command profile" can't contain
"metadata profile" and vice versa! This is strictly checked
and if the settings are mixed, such profile is rejected and
it's not used. So in the end, the CONFIG_PROFILE_COMMAND
set of options and CONFIG_PROFILE_METADATA are mutually exclusive
sets.
- Marking configuration with one or the other marker will also
determine the way these configuration sources are positioned
in the configuration cascade which is now:
CONFIG_STRING -> CONFIG_PROFILE_COMMAND -> CONFIG_PROFILE_METADATA -> CONFIG_FILE/CONFIG_MERGED_FILES
- Marking configuration with one or the other marker will also make
it possible to issue a command context refresh (will be probably
a part of a future patch) if needed for settings in global profile
set. For settings in metadata profile set this is impossible since
we can't refresh cmd context in the middle of reading VG/LV metadata
and for each VG/LV separately because each VG/LV can have a different
metadata profile assinged and it's not possible to change these
settings at this level.
- When command profile is incorrect, it's rejected *and also* the
command exits immediately - the profile *must* be correct for the
command that was run with a profile to be executed. Before this
patch, when the profile was found incorrect, there was just the
warning message and the command continued without profile applied.
But it's more correct to exit immediately in this case.
- When metadata profile is incorrect, we reject it during command
runtime (as we know the profile name from metadata and not early
from command line as it is in case of command profiles) and we
*do continue* with the command as we're in the middle of operation.
Also, the metadata profile is applied directly and on the fly on
find_config_tree_* fn call and even if the metadata profile is
found incorrect, we still need to return the non-profiled value
as found in the other configuration provided or default value.
To exit immediately even in this case, we'd need to refactor
existing find_config_tree_* fns so they can return error. Currently,
these fns return only config values (which end up with default
values in the end if the config is not found).
- To check the profile validity before use to be sure it's correct,
one can use :
lvm dumpconfig --commandprofile/--metadataprofile ProfileName --validate
(the --commandprofile/--metadataprofile for dumpconfig will come
as part of the subsequent patch)
- This patch also adds a reference to --commandprofile and
--metadataprofile in the cmd help string (which was missing before
for the --profile for some commands). We do not mention --profile
now as people should use --commandprofile or --metadataprofile
directly. However, the --profile is still supported for backward
compatibility and it's translated as:
--profile == --metadataprofile for lvcreate, vgcreate, lvchange and vgchange
(as these commands are able to attach profile to metadata)
--profile == --commandprofile for all the other commands
(--metadataprofile is not allowed there as it makes no sense)
- This patch also contains some cleanups to make the code handling
the profiles more readable...
2014-05-20 16:13:10 +04:00
if ( arg_count ( lv - > vg - > cmd , metadataprofile_ARG ) )
new_profile_name = arg_str_value ( lv - > vg - > cmd , metadataprofile_ARG , NULL ) ;
else
new_profile_name = arg_str_value ( lv - > vg - > cmd , profile_ARG , NULL ) ;
if ( ! ( new_profile = add_profile ( lv - > vg - > cmd , new_profile_name , CONFIG_PROFILE_METADATA ) ) )
2013-06-25 14:33:06 +04:00
return_0 ;
lv - > profile = new_profile ;
}
log_verbose ( " Changing configuration profile for LV %s: %s -> %s. " ,
lv - > name , old_profile_name , new_profile_name ) ;
if ( ! vg_write ( lv - > vg ) | | ! vg_commit ( lv - > vg ) )
return_0 ;
2013-07-11 14:25:10 +04:00
backup ( lv - > vg ) ;
2013-06-25 14:33:06 +04:00
return 1 ;
}
2015-02-26 21:38:26 +03:00
static int _lvchange_activation_skip ( struct logical_volume * lv )
2013-07-11 14:44:36 +04:00
{
int skip = arg_int_value ( lv - > vg - > cmd , setactivationskip_ARG , 0 ) ;
lv_set_activation_skip ( lv , 1 , skip ) ;
log_verbose ( " Changing activation skip flag to %s for LV %s. " ,
lv - > name , skip ? " enabled " : " disabled " ) ;
if ( ! vg_write ( lv - > vg ) | | ! vg_commit ( lv - > vg ) )
return_0 ;
backup ( lv - > vg ) ;
return 1 ;
}
2013-06-25 14:33:06 +04:00
2014-04-01 22:08:08 +04:00
static int _lvchange_single ( struct cmd_context * cmd , struct logical_volume * lv ,
2015-02-26 21:38:26 +03:00
struct processing_handle * handle __attribute__ ( ( unused ) ) )
2002-11-18 17:04:08 +03:00
{
2008-07-31 17:03:01 +04:00
int doit = 0 , docmds = 0 ;
2009-05-27 22:19:21 +04:00
struct logical_volume * origin ;
2011-11-18 23:22:49 +04:00
char snaps_msg [ 128 ] ;
2002-11-18 17:04:08 +03:00
if ( ! ( lv - > vg - > status & LVM_WRITE ) & &
( arg_count ( cmd , contiguous_ARG ) | | arg_count ( cmd , permission_ARG ) | |
2004-05-19 02:12:53 +04:00
arg_count ( cmd , readahead_ARG ) | | arg_count ( cmd , persistent_ARG ) | |
2013-06-25 14:33:06 +04:00
arg_count ( cmd , discards_ARG ) | | arg_count ( cmd , zero_ARG ) | |
config: differentiate command and metadata profiles and consolidate profile handling code
- When defining configuration source, the code now uses separate
CONFIG_PROFILE_COMMAND and CONFIG_PROFILE_METADATA markers
(before, it was just CONFIG_PROFILE that did not make the
difference between the two). This helps when checking the
configuration if it contains correct set of options which
are all in either command-profilable or metadata-profilable
group without mixing these groups together - so it's a firm
distinction. The "command profile" can't contain
"metadata profile" and vice versa! This is strictly checked
and if the settings are mixed, such profile is rejected and
it's not used. So in the end, the CONFIG_PROFILE_COMMAND
set of options and CONFIG_PROFILE_METADATA are mutually exclusive
sets.
- Marking configuration with one or the other marker will also
determine the way these configuration sources are positioned
in the configuration cascade which is now:
CONFIG_STRING -> CONFIG_PROFILE_COMMAND -> CONFIG_PROFILE_METADATA -> CONFIG_FILE/CONFIG_MERGED_FILES
- Marking configuration with one or the other marker will also make
it possible to issue a command context refresh (will be probably
a part of a future patch) if needed for settings in global profile
set. For settings in metadata profile set this is impossible since
we can't refresh cmd context in the middle of reading VG/LV metadata
and for each VG/LV separately because each VG/LV can have a different
metadata profile assinged and it's not possible to change these
settings at this level.
- When command profile is incorrect, it's rejected *and also* the
command exits immediately - the profile *must* be correct for the
command that was run with a profile to be executed. Before this
patch, when the profile was found incorrect, there was just the
warning message and the command continued without profile applied.
But it's more correct to exit immediately in this case.
- When metadata profile is incorrect, we reject it during command
runtime (as we know the profile name from metadata and not early
from command line as it is in case of command profiles) and we
*do continue* with the command as we're in the middle of operation.
Also, the metadata profile is applied directly and on the fly on
find_config_tree_* fn call and even if the metadata profile is
found incorrect, we still need to return the non-profiled value
as found in the other configuration provided or default value.
To exit immediately even in this case, we'd need to refactor
existing find_config_tree_* fns so they can return error. Currently,
these fns return only config values (which end up with default
values in the end if the config is not found).
- To check the profile validity before use to be sure it's correct,
one can use :
lvm dumpconfig --commandprofile/--metadataprofile ProfileName --validate
(the --commandprofile/--metadataprofile for dumpconfig will come
as part of the subsequent patch)
- This patch also adds a reference to --commandprofile and
--metadataprofile in the cmd help string (which was missing before
for the --profile for some commands). We do not mention --profile
now as people should use --commandprofile or --metadataprofile
directly. However, the --profile is still supported for backward
compatibility and it's translated as:
--profile == --metadataprofile for lvcreate, vgcreate, lvchange and vgchange
(as these commands are able to attach profile to metadata)
--profile == --commandprofile for all the other commands
(--metadataprofile is not allowed there as it makes no sense)
- This patch also contains some cleanups to make the code handling
the profiles more readable...
2014-05-20 16:13:10 +04:00
arg_count ( cmd , alloc_ARG ) | | arg_count ( cmd , profile_ARG ) | |
arg_count ( cmd , metadataprofile_ARG ) ) ) {
2002-11-18 17:04:08 +03:00
log_error ( " Only -a permitted with read-only volume "
" group \" %s \" " , lv - > vg - > name ) ;
return EINVALID_CMD_LINE ;
}
2012-09-19 16:55:34 +04:00
if ( lv_is_origin ( lv ) & & ! lv_is_thin_volume ( lv ) & &
2002-11-18 17:04:08 +03:00
( arg_count ( cmd , contiguous_ARG ) | | arg_count ( cmd , permission_ARG ) | |
2004-05-19 02:12:53 +04:00
arg_count ( cmd , readahead_ARG ) | | arg_count ( cmd , persistent_ARG ) | |
config: differentiate command and metadata profiles and consolidate profile handling code
- When defining configuration source, the code now uses separate
CONFIG_PROFILE_COMMAND and CONFIG_PROFILE_METADATA markers
(before, it was just CONFIG_PROFILE that did not make the
difference between the two). This helps when checking the
configuration if it contains correct set of options which
are all in either command-profilable or metadata-profilable
group without mixing these groups together - so it's a firm
distinction. The "command profile" can't contain
"metadata profile" and vice versa! This is strictly checked
and if the settings are mixed, such profile is rejected and
it's not used. So in the end, the CONFIG_PROFILE_COMMAND
set of options and CONFIG_PROFILE_METADATA are mutually exclusive
sets.
- Marking configuration with one or the other marker will also
determine the way these configuration sources are positioned
in the configuration cascade which is now:
CONFIG_STRING -> CONFIG_PROFILE_COMMAND -> CONFIG_PROFILE_METADATA -> CONFIG_FILE/CONFIG_MERGED_FILES
- Marking configuration with one or the other marker will also make
it possible to issue a command context refresh (will be probably
a part of a future patch) if needed for settings in global profile
set. For settings in metadata profile set this is impossible since
we can't refresh cmd context in the middle of reading VG/LV metadata
and for each VG/LV separately because each VG/LV can have a different
metadata profile assinged and it's not possible to change these
settings at this level.
- When command profile is incorrect, it's rejected *and also* the
command exits immediately - the profile *must* be correct for the
command that was run with a profile to be executed. Before this
patch, when the profile was found incorrect, there was just the
warning message and the command continued without profile applied.
But it's more correct to exit immediately in this case.
- When metadata profile is incorrect, we reject it during command
runtime (as we know the profile name from metadata and not early
from command line as it is in case of command profiles) and we
*do continue* with the command as we're in the middle of operation.
Also, the metadata profile is applied directly and on the fly on
find_config_tree_* fn call and even if the metadata profile is
found incorrect, we still need to return the non-profiled value
as found in the other configuration provided or default value.
To exit immediately even in this case, we'd need to refactor
existing find_config_tree_* fns so they can return error. Currently,
these fns return only config values (which end up with default
values in the end if the config is not found).
- To check the profile validity before use to be sure it's correct,
one can use :
lvm dumpconfig --commandprofile/--metadataprofile ProfileName --validate
(the --commandprofile/--metadataprofile for dumpconfig will come
as part of the subsequent patch)
- This patch also adds a reference to --commandprofile and
--metadataprofile in the cmd help string (which was missing before
for the --profile for some commands). We do not mention --profile
now as people should use --commandprofile or --metadataprofile
directly. However, the --profile is still supported for backward
compatibility and it's translated as:
--profile == --metadataprofile for lvcreate, vgcreate, lvchange and vgchange
(as these commands are able to attach profile to metadata)
--profile == --commandprofile for all the other commands
(--metadataprofile is not allowed there as it makes no sense)
- This patch also contains some cleanups to make the code handling
the profiles more readable...
2014-05-20 16:13:10 +04:00
arg_count ( cmd , alloc_ARG ) | | arg_count ( cmd , profile_ARG ) | |
arg_count ( cmd , metadataprofile_ARG ) ) ) {
2002-11-18 17:04:08 +03:00
log_error ( " Can't change logical volume \" %s \" under snapshot " ,
lv - > name ) ;
return ECMD_FAILED ;
}
2014-09-16 00:33:53 +04:00
if ( lv_is_pvmove ( lv ) ) {
2003-05-06 16:14:36 +04:00
log_error ( " Unable to change pvmove LV %s " , lv - > name ) ;
2012-06-27 15:48:31 +04:00
if ( arg_count ( cmd , activate_ARG ) )
2003-05-06 16:14:36 +04:00
log_error ( " Use 'pvmove --abort' to abandon a pvmove " ) ;
return ECMD_FAILED ;
}
2014-09-16 00:33:53 +04:00
if ( lv_is_mirror_log ( lv ) ) {
2005-06-01 20:51:55 +04:00
log_error ( " Unable to change mirror log LV %s directly " , lv - > name ) ;
return ECMD_FAILED ;
}
2014-09-16 00:33:53 +04:00
if ( lv_is_mirror_image ( lv ) ) {
2005-06-03 18:49:51 +04:00
log_error ( " Unable to change mirror image LV %s directly " ,
lv - > name ) ;
return ECMD_FAILED ;
}
2009-05-27 22:19:21 +04:00
/* If LV is sparse, activate origin instead */
2012-06-27 15:48:31 +04:00
if ( arg_count ( cmd , activate_ARG ) & & lv_is_cow ( lv ) & &
2009-05-27 22:19:21 +04:00
lv_is_virtual_origin ( origin = origin_from_cow ( lv ) ) )
lv = origin ;
2014-10-13 18:21:41 +04:00
if ( ( lv_is_thin_pool_data ( lv ) | | lv_is_thin_pool_metadata ( lv ) | |
lv_is_cache_pool_data ( lv ) | | lv_is_cache_pool_metadata ( lv ) ) & &
2013-08-10 03:13:27 +04:00
! arg_count ( cmd , activate_ARG ) & &
! arg_count ( cmd , permission_ARG ) & &
! arg_count ( cmd , setactivationskip_ARG ) )
/* Rest can be changed for stacked thin pool meta/data volumes */
;
2014-08-15 15:52:21 +04:00
else if ( ! lv_is_visible ( lv ) & & ! lv_is_virtual_origin ( lv ) ) {
2005-06-03 18:49:51 +04:00
log_error ( " Unable to change internal LV %s directly " ,
lv - > name ) ;
return ECMD_FAILED ;
}
2014-08-15 15:31:53 +04:00
if ( lv_is_cow ( lv ) & & arg_count ( cmd , activate_ARG ) ) {
2014-08-15 17:51:03 +04:00
origin = origin_from_cow ( lv ) ;
2014-08-15 15:31:53 +04:00
if ( origin - > origin_count < 2 )
snaps_msg [ 0 ] = ' \0 ' ;
else if ( dm_snprintf ( snaps_msg , sizeof ( snaps_msg ) ,
" and %u other snapshot(s) " ,
origin - > origin_count - 1 ) < 0 ) {
log_error ( " Failed to prepare message. " ) ;
return ECMD_FAILED ;
}
if ( ! arg_count ( cmd , yes_ARG ) & &
( yes_no_prompt ( " Change of snapshot %s will also change its "
2014-08-15 17:52:01 +04:00
" origin %s%s. Proceed? [y/n]: " ,
display_lvname ( lv ) , display_lvname ( origin ) ,
snaps_msg ) = = ' n ' ) ) {
log_error ( " Logical volume %s not changed. " , display_lvname ( lv ) ) ;
2014-08-15 15:31:53 +04:00
return ECMD_FAILED ;
}
}
2015-01-15 17:20:08 +03:00
if ( arg_is_set ( cmd , errorwhenfull_ARG ) & & ! lv_is_thin_pool ( lv ) ) {
log_error ( " Option --errorwhenfull is only supported with thin pools. " ) ;
return ECMD_FAILED ;
}
2015-03-17 11:44:26 +03:00
if ( arg_count ( cmd , persistent_ARG ) & & lv_is_pool ( lv ) ) {
log_error ( " Persistent major and minor numbers are not supported with pools. " ) ;
return ECMD_FAILED ;
}
2015-03-05 23:00:44 +03:00
if ( ! arg_count ( cmd , activate_ARG ) & & ! arg_count ( cmd , refresh_ARG ) ) {
/*
* If a persistent lv lock already exists from activation
* ( with the needed mode or higher ) , this will be a no - op .
* Otherwise , the lv lock will be taken as non - persistent
* and released when this command exits .
*
* FIXME : use " sh " if the options imply that the lvchange
* operation does not modify the LV .
*/
if ( ! lockd_lv ( cmd , lv , " ex " , 0 ) ) {
stack ;
return ECMD_FAILED ;
}
}
2010-01-06 22:08:58 +03:00
/*
* FIXME : DEFAULT_BACKGROUND_POLLING should be " unspecified " .
* If - - poll is explicitly provided use it ; otherwise polling
* should only be started if the LV is not already active . So :
* 1 ) change the activation code to say if the LV was actually activated
* 2 ) make polling of an LV tightly coupled with LV activation
2010-05-06 15:15:55 +04:00
*
* Do not initiate any polling if - - sysinit option is used .
2010-01-06 22:08:58 +03:00
*/
2010-05-06 15:15:55 +04:00
init_background_polling ( arg_count ( cmd , sysinit_ARG ) ? 0 :
arg_int_value ( cmd , poll_ARG ,
DEFAULT_BACKGROUND_POLLING ) ) ;
2010-01-05 23:56:51 +03:00
2002-11-18 17:04:08 +03:00
/* access permission change */
if ( arg_count ( cmd , permission_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_permission ( cmd , lv ) ;
2008-07-31 17:03:01 +04:00
docmds + + ;
2002-11-18 17:04:08 +03:00
}
/* allocation policy change */
2004-05-19 02:12:53 +04:00
if ( arg_count ( cmd , contiguous_ARG ) | | arg_count ( cmd , alloc_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_alloc ( cmd , lv ) ;
2008-07-31 17:03:01 +04:00
docmds + + ;
2002-11-18 17:04:08 +03:00
}
2015-01-15 17:20:08 +03:00
/* error when full change */
if ( arg_count ( cmd , errorwhenfull_ARG ) ) {
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_errorwhenfull ( cmd , lv ) ;
2015-01-15 17:20:08 +03:00
docmds + + ;
}
2002-11-18 17:04:08 +03:00
/* read ahead sector change */
if ( arg_count ( cmd , readahead_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_readahead ( cmd , lv ) ;
2008-07-31 17:03:01 +04:00
docmds + + ;
2002-11-18 17:04:08 +03:00
}
2009-11-04 15:39:56 +03:00
/* persistent device number change */
2002-11-18 17:04:08 +03:00
if ( arg_count ( cmd , persistent_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_persistent ( cmd , lv ) ;
2008-07-31 17:03:01 +04:00
docmds + + ;
2013-07-01 13:27:22 +04:00
if ( sigint_caught ( ) )
return_ECMD_FAILED ;
2002-11-18 17:04:08 +03:00
}
2012-08-08 00:24:41 +04:00
if ( arg_count ( cmd , discards_ARG ) | |
2012-06-28 16:52:23 +04:00
arg_count ( cmd , zero_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_pool_update ( cmd , lv ) ;
2012-06-28 16:52:23 +04:00
docmds + + ;
}
2004-03-08 20:19:15 +03:00
/* add tag */
if ( arg_count ( cmd , addtag_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_tag ( cmd , lv , addtag_ARG ) ;
2008-07-31 17:03:01 +04:00
docmds + + ;
2004-03-08 20:19:15 +03:00
}
/* del tag */
if ( arg_count ( cmd , deltag_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_tag ( cmd , lv , deltag_ARG ) ;
2008-07-31 17:03:01 +04:00
docmds + + ;
2004-03-08 20:19:15 +03:00
}
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
/* change writemostly/writebehind */
if ( arg_count ( cmd , writemostly_ARG ) | | arg_count ( cmd , writebehind_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_writemostly ( lv ) ;
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
docmds + + ;
}
2013-05-31 20:25:52 +04:00
/* change [min|max]_recovery_rate */
if ( arg_count ( cmd , minrecoveryrate_ARG ) | |
arg_count ( cmd , maxrecoveryrate_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_recovery_rate ( lv ) ;
2013-05-31 20:25:52 +04:00
docmds + + ;
}
2013-06-25 14:33:06 +04:00
/* change configuration profile */
config: differentiate command and metadata profiles and consolidate profile handling code
- When defining configuration source, the code now uses separate
CONFIG_PROFILE_COMMAND and CONFIG_PROFILE_METADATA markers
(before, it was just CONFIG_PROFILE that did not make the
difference between the two). This helps when checking the
configuration if it contains correct set of options which
are all in either command-profilable or metadata-profilable
group without mixing these groups together - so it's a firm
distinction. The "command profile" can't contain
"metadata profile" and vice versa! This is strictly checked
and if the settings are mixed, such profile is rejected and
it's not used. So in the end, the CONFIG_PROFILE_COMMAND
set of options and CONFIG_PROFILE_METADATA are mutually exclusive
sets.
- Marking configuration with one or the other marker will also
determine the way these configuration sources are positioned
in the configuration cascade which is now:
CONFIG_STRING -> CONFIG_PROFILE_COMMAND -> CONFIG_PROFILE_METADATA -> CONFIG_FILE/CONFIG_MERGED_FILES
- Marking configuration with one or the other marker will also make
it possible to issue a command context refresh (will be probably
a part of a future patch) if needed for settings in global profile
set. For settings in metadata profile set this is impossible since
we can't refresh cmd context in the middle of reading VG/LV metadata
and for each VG/LV separately because each VG/LV can have a different
metadata profile assinged and it's not possible to change these
settings at this level.
- When command profile is incorrect, it's rejected *and also* the
command exits immediately - the profile *must* be correct for the
command that was run with a profile to be executed. Before this
patch, when the profile was found incorrect, there was just the
warning message and the command continued without profile applied.
But it's more correct to exit immediately in this case.
- When metadata profile is incorrect, we reject it during command
runtime (as we know the profile name from metadata and not early
from command line as it is in case of command profiles) and we
*do continue* with the command as we're in the middle of operation.
Also, the metadata profile is applied directly and on the fly on
find_config_tree_* fn call and even if the metadata profile is
found incorrect, we still need to return the non-profiled value
as found in the other configuration provided or default value.
To exit immediately even in this case, we'd need to refactor
existing find_config_tree_* fns so they can return error. Currently,
these fns return only config values (which end up with default
values in the end if the config is not found).
- To check the profile validity before use to be sure it's correct,
one can use :
lvm dumpconfig --commandprofile/--metadataprofile ProfileName --validate
(the --commandprofile/--metadataprofile for dumpconfig will come
as part of the subsequent patch)
- This patch also adds a reference to --commandprofile and
--metadataprofile in the cmd help string (which was missing before
for the --profile for some commands). We do not mention --profile
now as people should use --commandprofile or --metadataprofile
directly. However, the --profile is still supported for backward
compatibility and it's translated as:
--profile == --metadataprofile for lvcreate, vgcreate, lvchange and vgchange
(as these commands are able to attach profile to metadata)
--profile == --commandprofile for all the other commands
(--metadataprofile is not allowed there as it makes no sense)
- This patch also contains some cleanups to make the code handling
the profiles more readable...
2014-05-20 16:13:10 +04:00
if ( arg_count ( cmd , profile_ARG ) | | arg_count ( cmd , metadataprofile_ARG ) | |
arg_count ( cmd , detachprofile_ARG ) ) {
2013-06-25 14:33:06 +04:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_profile ( lv ) ;
2013-06-25 14:33:06 +04:00
docmds + + ;
}
2013-07-11 14:44:36 +04:00
if ( arg_count ( cmd , setactivationskip_ARG ) ) {
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_activation_skip ( lv ) ;
2013-07-11 14:44:36 +04:00
docmds + + ;
}
2014-11-20 13:30:19 +03:00
if ( arg_count ( cmd , cachepolicy_ARG ) | | arg_count ( cmd , cachesettings_ARG ) ) {
2014-11-19 20:39:58 +03:00
if ( ! archive ( lv - > vg ) )
return_ECMD_FAILED ;
2015-02-26 21:38:26 +03:00
doit + = _lvchange_cachepolicy ( cmd , lv ) ;
2014-11-19 20:39:58 +03:00
docmds + + ;
}
2002-11-18 17:04:08 +03:00
if ( doit )
2013-06-25 15:43:35 +04:00
log_print_unless_silent ( " Logical volume \" %s \" changed. " , lv - > name ) ;
2002-11-18 17:04:08 +03:00
2013-07-01 13:27:22 +04:00
if ( arg_count ( cmd , resync_ARG ) & &
2015-02-26 21:38:26 +03:00
! _lvchange_resync ( cmd , lv ) )
2013-07-01 13:27:22 +04:00
return_ECMD_FAILED ;
2006-10-24 03:03:55 +04:00
2013-07-01 13:27:22 +04:00
if ( arg_count ( cmd , syncaction_ARG ) & &
! lv_raid_message ( lv , arg_str_value ( cmd , syncaction_ARG , NULL ) ) )
return_ECMD_FAILED ;
2013-04-12 00:33:59 +04:00
2012-06-27 15:48:31 +04:00
/* activation change */
if ( arg_count ( cmd , activate_ARG ) ) {
2013-07-01 13:27:22 +04:00
if ( ! _lvchange_activate ( cmd , lv ) )
return_ECMD_FAILED ;
} else if ( arg_count ( cmd , refresh_ARG ) ) {
2015-02-26 21:38:26 +03:00
if ( ! _lvchange_refresh ( cmd , lv ) )
2013-07-01 13:27:22 +04:00
return_ECMD_FAILED ;
} else {
if ( arg_count ( cmd , monitor_ARG ) & &
2015-02-26 21:38:26 +03:00
! _lvchange_monitoring ( cmd , lv ) )
2013-07-01 13:27:22 +04:00
return_ECMD_FAILED ;
2006-05-12 23:16:48 +04:00
2013-07-01 13:27:22 +04:00
if ( arg_count ( cmd , poll_ARG ) & &
2015-02-26 21:38:26 +03:00
! _lvchange_background_polling ( cmd , lv ) )
2013-07-01 13:27:22 +04:00
return_ECMD_FAILED ;
2010-01-05 23:56:51 +03:00
}
2013-07-01 13:27:22 +04:00
if ( doit ! = docmds )
return_ECMD_FAILED ;
2008-07-31 17:03:01 +04:00
2003-10-22 02:06:07 +04:00
return ECMD_PROCESSED ;
2002-11-18 17:04:08 +03:00
}
int lvchange ( struct cmd_context * cmd , int argc , char * * argv )
{
[lv|vg]change: Allow limited metadata changes when PVs are missing
A while back, the behavior of LVM changed from allowing metadata changes
when PVs were missing to not allowing changes. Until recently, this
change was tolerated by HA-LVM by forcing a 'vgreduce --removemissing'
before trying (again) to add tags to an LV and then activate it. LVM
mirroring requires that failed devices are removed anyway, so this was
largely harmless. However, RAID LVs do not require devices to be removed
from the array in order to be activated. In fact, in an HA-LVM
environment this would be very undesirable. Device failures in such an
environment can often be transient and it would be much better to restore
the device to the array than synchronize an entirely new device.
There are two methods that can be used to setup an HA-LVM environment:
"clvm" or "tagging". For RAID LVs, "clvm" is out of the question because
RAID LVs are not supported in clustered VGs - not even in an exclusively
activated manner. That leaves "tagging". HA-LVM uses tagging - coupled
with 'volume_list' - to ensure that only one machine can have an LV active
at a time. If updates are not allowed when a PV is missing, it is
impossible to add or remove tags to allow for activation. This removes
one of the most basic functionalities of HA-LVM - site redundancy. If
mirroring or RAID is used to replicate the storage in two data centers
and one of them goes down, a server and a storage device are lost. When
the service fails-over to the alternate site, the VG will be "partial".
Unable to add a tag to the VG/LV, the RAID device will be unable to
activate.
The solution is to allow vgchange and lvchange to alter the LVM metadata
for a limited set of options - --[add|del]tag included. The set of
allowable options are ones that do not cause changes to the DM kernel
target (like --resync would) or could alter the structure of the LV
(like allocation or conversion).
2012-10-10 20:33:10 +04:00
/*
* Options that update metadata should be listed in one of
* the two lists below ( i . e . options other than - a , - - refresh ,
* - - monitor or - - poll ) .
*/
int update_partial_safe = /* options safe to update if partial */
arg_count ( cmd , contiguous_ARG ) | |
arg_count ( cmd , permission_ARG ) | |
arg_count ( cmd , readahead_ARG ) | |
arg_count ( cmd , persistent_ARG ) | |
arg_count ( cmd , addtag_ARG ) | |
2013-06-25 14:33:06 +04:00
arg_count ( cmd , deltag_ARG ) | |
config: differentiate command and metadata profiles and consolidate profile handling code
- When defining configuration source, the code now uses separate
CONFIG_PROFILE_COMMAND and CONFIG_PROFILE_METADATA markers
(before, it was just CONFIG_PROFILE that did not make the
difference between the two). This helps when checking the
configuration if it contains correct set of options which
are all in either command-profilable or metadata-profilable
group without mixing these groups together - so it's a firm
distinction. The "command profile" can't contain
"metadata profile" and vice versa! This is strictly checked
and if the settings are mixed, such profile is rejected and
it's not used. So in the end, the CONFIG_PROFILE_COMMAND
set of options and CONFIG_PROFILE_METADATA are mutually exclusive
sets.
- Marking configuration with one or the other marker will also
determine the way these configuration sources are positioned
in the configuration cascade which is now:
CONFIG_STRING -> CONFIG_PROFILE_COMMAND -> CONFIG_PROFILE_METADATA -> CONFIG_FILE/CONFIG_MERGED_FILES
- Marking configuration with one or the other marker will also make
it possible to issue a command context refresh (will be probably
a part of a future patch) if needed for settings in global profile
set. For settings in metadata profile set this is impossible since
we can't refresh cmd context in the middle of reading VG/LV metadata
and for each VG/LV separately because each VG/LV can have a different
metadata profile assinged and it's not possible to change these
settings at this level.
- When command profile is incorrect, it's rejected *and also* the
command exits immediately - the profile *must* be correct for the
command that was run with a profile to be executed. Before this
patch, when the profile was found incorrect, there was just the
warning message and the command continued without profile applied.
But it's more correct to exit immediately in this case.
- When metadata profile is incorrect, we reject it during command
runtime (as we know the profile name from metadata and not early
from command line as it is in case of command profiles) and we
*do continue* with the command as we're in the middle of operation.
Also, the metadata profile is applied directly and on the fly on
find_config_tree_* fn call and even if the metadata profile is
found incorrect, we still need to return the non-profiled value
as found in the other configuration provided or default value.
To exit immediately even in this case, we'd need to refactor
existing find_config_tree_* fns so they can return error. Currently,
these fns return only config values (which end up with default
values in the end if the config is not found).
- To check the profile validity before use to be sure it's correct,
one can use :
lvm dumpconfig --commandprofile/--metadataprofile ProfileName --validate
(the --commandprofile/--metadataprofile for dumpconfig will come
as part of the subsequent patch)
- This patch also adds a reference to --commandprofile and
--metadataprofile in the cmd help string (which was missing before
for the --profile for some commands). We do not mention --profile
now as people should use --commandprofile or --metadataprofile
directly. However, the --profile is still supported for backward
compatibility and it's translated as:
--profile == --metadataprofile for lvcreate, vgcreate, lvchange and vgchange
(as these commands are able to attach profile to metadata)
--profile == --commandprofile for all the other commands
(--metadataprofile is not allowed there as it makes no sense)
- This patch also contains some cleanups to make the code handling
the profiles more readable...
2014-05-20 16:13:10 +04:00
arg_count ( cmd , metadataprofile_ARG ) | |
2013-06-25 14:33:06 +04:00
arg_count ( cmd , profile_ARG ) | |
2013-07-11 14:44:36 +04:00
arg_count ( cmd , detachprofile_ARG ) | |
arg_count ( cmd , setactivationskip_ARG ) ;
[lv|vg]change: Allow limited metadata changes when PVs are missing
A while back, the behavior of LVM changed from allowing metadata changes
when PVs were missing to not allowing changes. Until recently, this
change was tolerated by HA-LVM by forcing a 'vgreduce --removemissing'
before trying (again) to add tags to an LV and then activate it. LVM
mirroring requires that failed devices are removed anyway, so this was
largely harmless. However, RAID LVs do not require devices to be removed
from the array in order to be activated. In fact, in an HA-LVM
environment this would be very undesirable. Device failures in such an
environment can often be transient and it would be much better to restore
the device to the array than synchronize an entirely new device.
There are two methods that can be used to setup an HA-LVM environment:
"clvm" or "tagging". For RAID LVs, "clvm" is out of the question because
RAID LVs are not supported in clustered VGs - not even in an exclusively
activated manner. That leaves "tagging". HA-LVM uses tagging - coupled
with 'volume_list' - to ensure that only one machine can have an LV active
at a time. If updates are not allowed when a PV is missing, it is
impossible to add or remove tags to allow for activation. This removes
one of the most basic functionalities of HA-LVM - site redundancy. If
mirroring or RAID is used to replicate the storage in two data centers
and one of them goes down, a server and a storage device are lost. When
the service fails-over to the alternate site, the VG will be "partial".
Unable to add a tag to the VG/LV, the RAID device will be unable to
activate.
The solution is to allow vgchange and lvchange to alter the LVM metadata
for a limited set of options - --[add|del]tag included. The set of
allowable options are ones that do not cause changes to the DM kernel
target (like --resync would) or could alter the structure of the LV
(like allocation or conversion).
2012-10-10 20:33:10 +04:00
int update_partial_unsafe =
arg_count ( cmd , alloc_ARG ) | |
arg_count ( cmd , discards_ARG ) | |
2015-01-15 17:20:08 +03:00
arg_count ( cmd , errorwhenfull_ARG ) | |
2013-08-12 21:40:52 +04:00
arg_count ( cmd , minrecoveryrate_ARG ) | |
arg_count ( cmd , maxrecoveryrate_ARG ) | |
2013-08-10 02:09:47 +04:00
arg_count ( cmd , resync_ARG ) | |
2013-04-12 00:33:59 +04:00
arg_count ( cmd , syncaction_ARG ) | |
2014-11-19 20:39:58 +03:00
arg_count ( cmd , cachepolicy_ARG ) | |
2014-11-20 13:30:19 +03:00
arg_count ( cmd , cachesettings_ARG ) | |
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
arg_count ( cmd , writebehind_ARG ) | |
arg_count ( cmd , writemostly_ARG ) | |
[lv|vg]change: Allow limited metadata changes when PVs are missing
A while back, the behavior of LVM changed from allowing metadata changes
when PVs were missing to not allowing changes. Until recently, this
change was tolerated by HA-LVM by forcing a 'vgreduce --removemissing'
before trying (again) to add tags to an LV and then activate it. LVM
mirroring requires that failed devices are removed anyway, so this was
largely harmless. However, RAID LVs do not require devices to be removed
from the array in order to be activated. In fact, in an HA-LVM
environment this would be very undesirable. Device failures in such an
environment can often be transient and it would be much better to restore
the device to the array than synchronize an entirely new device.
There are two methods that can be used to setup an HA-LVM environment:
"clvm" or "tagging". For RAID LVs, "clvm" is out of the question because
RAID LVs are not supported in clustered VGs - not even in an exclusively
activated manner. That leaves "tagging". HA-LVM uses tagging - coupled
with 'volume_list' - to ensure that only one machine can have an LV active
at a time. If updates are not allowed when a PV is missing, it is
impossible to add or remove tags to allow for activation. This removes
one of the most basic functionalities of HA-LVM - site redundancy. If
mirroring or RAID is used to replicate the storage in two data centers
and one of them goes down, a server and a storage device are lost. When
the service fails-over to the alternate site, the VG will be "partial".
Unable to add a tag to the VG/LV, the RAID device will be unable to
activate.
The solution is to allow vgchange and lvchange to alter the LVM metadata
for a limited set of options - --[add|del]tag included. The set of
allowable options are ones that do not cause changes to the DM kernel
target (like --resync would) or could alter the structure of the LV
(like allocation or conversion).
2012-10-10 20:33:10 +04:00
arg_count ( cmd , zero_ARG ) ;
int update = update_partial_safe | | update_partial_unsafe ;
2010-06-28 23:10:16 +04:00
if ( ! update & &
2015-02-26 21:38:26 +03:00
! arg_count ( cmd , activate_ARG ) & & ! arg_count ( cmd , refresh_ARG ) & &
! arg_count ( cmd , monitor_ARG ) & & ! arg_count ( cmd , poll_ARG ) ) {
2012-08-07 23:08:14 +04:00
log_error ( " Need 1 or more of -a, -C, -M, -p, -r, -Z, "
2010-01-05 23:56:51 +03:00
" --resync, --refresh, --alloc, --addtag, --deltag, "
2012-08-08 00:24:41 +04:00
" --monitor, --poll or --discards " ) ;
2002-11-18 17:04:08 +03:00
return EINVALID_CMD_LINE ;
}
config: differentiate command and metadata profiles and consolidate profile handling code
- When defining configuration source, the code now uses separate
CONFIG_PROFILE_COMMAND and CONFIG_PROFILE_METADATA markers
(before, it was just CONFIG_PROFILE that did not make the
difference between the two). This helps when checking the
configuration if it contains correct set of options which
are all in either command-profilable or metadata-profilable
group without mixing these groups together - so it's a firm
distinction. The "command profile" can't contain
"metadata profile" and vice versa! This is strictly checked
and if the settings are mixed, such profile is rejected and
it's not used. So in the end, the CONFIG_PROFILE_COMMAND
set of options and CONFIG_PROFILE_METADATA are mutually exclusive
sets.
- Marking configuration with one or the other marker will also
determine the way these configuration sources are positioned
in the configuration cascade which is now:
CONFIG_STRING -> CONFIG_PROFILE_COMMAND -> CONFIG_PROFILE_METADATA -> CONFIG_FILE/CONFIG_MERGED_FILES
- Marking configuration with one or the other marker will also make
it possible to issue a command context refresh (will be probably
a part of a future patch) if needed for settings in global profile
set. For settings in metadata profile set this is impossible since
we can't refresh cmd context in the middle of reading VG/LV metadata
and for each VG/LV separately because each VG/LV can have a different
metadata profile assinged and it's not possible to change these
settings at this level.
- When command profile is incorrect, it's rejected *and also* the
command exits immediately - the profile *must* be correct for the
command that was run with a profile to be executed. Before this
patch, when the profile was found incorrect, there was just the
warning message and the command continued without profile applied.
But it's more correct to exit immediately in this case.
- When metadata profile is incorrect, we reject it during command
runtime (as we know the profile name from metadata and not early
from command line as it is in case of command profiles) and we
*do continue* with the command as we're in the middle of operation.
Also, the metadata profile is applied directly and on the fly on
find_config_tree_* fn call and even if the metadata profile is
found incorrect, we still need to return the non-profiled value
as found in the other configuration provided or default value.
To exit immediately even in this case, we'd need to refactor
existing find_config_tree_* fns so they can return error. Currently,
these fns return only config values (which end up with default
values in the end if the config is not found).
- To check the profile validity before use to be sure it's correct,
one can use :
lvm dumpconfig --commandprofile/--metadataprofile ProfileName --validate
(the --commandprofile/--metadataprofile for dumpconfig will come
as part of the subsequent patch)
- This patch also adds a reference to --commandprofile and
--metadataprofile in the cmd help string (which was missing before
for the --profile for some commands). We do not mention --profile
now as people should use --commandprofile or --metadataprofile
directly. However, the --profile is still supported for backward
compatibility and it's translated as:
--profile == --metadataprofile for lvcreate, vgcreate, lvchange and vgchange
(as these commands are able to attach profile to metadata)
--profile == --commandprofile for all the other commands
(--metadataprofile is not allowed there as it makes no sense)
- This patch also contains some cleanups to make the code handling
the profiles more readable...
2014-05-20 16:13:10 +04:00
if ( ( arg_count ( cmd , profile_ARG ) | | arg_count ( cmd , metadataprofile_ARG ) ) & &
arg_count ( cmd , detachprofile_ARG ) ) {
log_error ( " Only one of --metadataprofile and --detachprofile permitted. " ) ;
2013-06-25 14:33:06 +04:00
return EINVALID_CMD_LINE ;
}
2012-06-27 15:48:31 +04:00
if ( arg_count ( cmd , activate_ARG ) & & arg_count ( cmd , refresh_ARG ) ) {
2010-10-26 05:37:59 +04:00
log_error ( " Only one of -a and --refresh permitted. " ) ;
return EINVALID_CMD_LINE ;
}
2009-07-15 09:47:55 +04:00
2010-05-06 15:15:55 +04:00
if ( ( arg_count ( cmd , ignorelockingfailure_ARG ) | |
2010-06-28 23:10:16 +04:00
arg_count ( cmd , sysinit_ARG ) ) & & update ) {
2010-05-06 15:15:55 +04:00
log_error ( " Only -a permitted with --ignorelockingfailure and --sysinit " ) ;
2002-11-18 17:04:08 +03:00
return EINVALID_CMD_LINE ;
}
[lv|vg]change: Allow limited metadata changes when PVs are missing
A while back, the behavior of LVM changed from allowing metadata changes
when PVs were missing to not allowing changes. Until recently, this
change was tolerated by HA-LVM by forcing a 'vgreduce --removemissing'
before trying (again) to add tags to an LV and then activate it. LVM
mirroring requires that failed devices are removed anyway, so this was
largely harmless. However, RAID LVs do not require devices to be removed
from the array in order to be activated. In fact, in an HA-LVM
environment this would be very undesirable. Device failures in such an
environment can often be transient and it would be much better to restore
the device to the array than synchronize an entirely new device.
There are two methods that can be used to setup an HA-LVM environment:
"clvm" or "tagging". For RAID LVs, "clvm" is out of the question because
RAID LVs are not supported in clustered VGs - not even in an exclusively
activated manner. That leaves "tagging". HA-LVM uses tagging - coupled
with 'volume_list' - to ensure that only one machine can have an LV active
at a time. If updates are not allowed when a PV is missing, it is
impossible to add or remove tags to allow for activation. This removes
one of the most basic functionalities of HA-LVM - site redundancy. If
mirroring or RAID is used to replicate the storage in two data centers
and one of them goes down, a server and a storage device are lost. When
the service fails-over to the alternate site, the VG will be "partial".
Unable to add a tag to the VG/LV, the RAID device will be unable to
activate.
The solution is to allow vgchange and lvchange to alter the LVM metadata
for a limited set of options - --[add|del]tag included. The set of
allowable options are ones that do not cause changes to the DM kernel
target (like --resync would) or could alter the structure of the LV
(like allocation or conversion).
2012-10-10 20:33:10 +04:00
if ( ! update | | ! update_partial_unsafe )
2009-07-15 09:47:55 +04:00
cmd - > handles_missing_pvs = 1 ;
2014-12-03 12:03:48 +03:00
if ( ! argc & & ! arg_is_set ( cmd , select_ARG ) ) {
2015-03-04 16:40:04 +03:00
log_error ( " Please give logical volume path(s) or use --select for selection. " ) ;
2002-11-18 17:04:08 +03:00
return EINVALID_CMD_LINE ;
}
2003-04-02 23:14:43 +04:00
if ( ( arg_count ( cmd , minor_ARG ) | | arg_count ( cmd , major_ARG ) ) & &
! arg_count ( cmd , persistent_ARG ) ) {
log_error ( " --major and --minor require -My " ) ;
return EINVALID_CMD_LINE ;
}
2002-11-18 17:04:08 +03:00
if ( arg_count ( cmd , minor_ARG ) & & argc ! = 1 ) {
log_error ( " Only give one logical volume when specifying minor " ) ;
return EINVALID_CMD_LINE ;
}
2004-05-19 02:12:53 +04:00
if ( arg_count ( cmd , contiguous_ARG ) & & arg_count ( cmd , alloc_ARG ) ) {
log_error ( " Only one of --alloc and --contiguous permitted " ) ;
return EINVALID_CMD_LINE ;
}
2010-05-06 15:15:55 +04:00
if ( arg_count ( cmd , poll_ARG ) & & arg_count ( cmd , sysinit_ARG ) ) {
log_error ( " Only one of --poll and --sysinit permitted " ) ;
return EINVALID_CMD_LINE ;
}
2013-11-26 17:51:23 +04:00
/*
* If - - sysinit - aay is used and at the same time lvmetad is used ,
* we want to rely on autoactivation to take place . Also , we
* need to take special care here as lvmetad service does
* not neet to be running at this moment yet - it could be
* just too early during system initialization time .
*/
if ( arg_count ( cmd , sysinit_ARG ) & & lvmetad_used ( ) & &
2012-07-10 15:49:46 +04:00
arg_uint_value ( cmd , activate_ARG , 0 ) = = CHANGE_AAY ) {
2013-11-26 17:51:23 +04:00
if ( ! lvmetad_socket_present ( ) ) {
/*
* If lvmetad socket is not present yet ,
* the service is just not started . It ' ll
* be started a bit later so we need to do
* the activation without lvmetad which means
* direct activation instead of autoactivation .
*/
log_warn ( " lvmetad is not active yet, using direct activation during sysinit " ) ;
2014-09-30 18:08:05 +04:00
lvmetad_set_active ( cmd , 0 ) ;
2013-11-26 17:51:23 +04:00
} else if ( lvmetad_active ( ) ) {
/*
* If lvmetad is active already , we want
* to make use of the autoactivation .
*/
log_warn ( " lvmetad is active, skipping direct activation during sysinit " ) ;
return ECMD_PROCESSED ;
}
2012-07-10 15:49:46 +04:00
}
2015-03-05 23:00:44 +03:00
/*
* Include foreign VGs that contain active LVs .
* That shouldn ' t happen in general , but if it does by some
* mistake , then we want to allow those LVs to be deactivated .
*/
2015-02-25 20:33:11 +03:00
if ( arg_is_set ( cmd , activate_ARG ) )
cmd - > include_active_foreign_vgs = 1 ;
2015-03-05 23:00:44 +03:00
/*
* The default vg lock mode for lvchange is ex , but these options
* are cases where lvchange does not modify the vg , so they can use
* the sh lock mode .
*/
vgchange/lvchange: enforce the shared VG lock from lvmlockd
The vgchange/lvchange activation commands read the VG, and
don't write it, so they acquire a shared VG lock from lvmlockd.
When other commands fail to acquire a shared VG lock from
lvmlockd, a warning is printed and they continue without it.
(Without it, the VG metadata they display from lvmetad may
not be up to date.)
vgchange/lvchange -a shouldn't continue without the shared
lock for a couple reasons:
. Usually they will just continue on and fail to acquire the
LV locks for activation, so continuing is pointless.
. More importantly, without the sh VG lock, the VG metadata
used by the command may be stale, and the LV locks shown
in the VG metadata may no longer be current. In the
case of sanlock, this would result in odd, unpredictable
errors when lvmlockd doesn't find the expected lock on
disk. In the case of dlm, the invalid LV lock could be
granted for the non-existing LV.
The solution is to not continue after the shared lock fails,
in the same way that a command fails if an exclusive lock fails.
2015-07-17 23:13:22 +03:00
if ( arg_count ( cmd , activate_ARG ) | | arg_count ( cmd , refresh_ARG ) ) {
2015-03-05 23:00:44 +03:00
cmd - > lockd_vg_default_sh = 1 ;
vgchange/lvchange: enforce the shared VG lock from lvmlockd
The vgchange/lvchange activation commands read the VG, and
don't write it, so they acquire a shared VG lock from lvmlockd.
When other commands fail to acquire a shared VG lock from
lvmlockd, a warning is printed and they continue without it.
(Without it, the VG metadata they display from lvmetad may
not be up to date.)
vgchange/lvchange -a shouldn't continue without the shared
lock for a couple reasons:
. Usually they will just continue on and fail to acquire the
LV locks for activation, so continuing is pointless.
. More importantly, without the sh VG lock, the VG metadata
used by the command may be stale, and the LV locks shown
in the VG metadata may no longer be current. In the
case of sanlock, this would result in odd, unpredictable
errors when lvmlockd doesn't find the expected lock on
disk. In the case of dlm, the invalid LV lock could be
granted for the non-existing LV.
The solution is to not continue after the shared lock fails,
in the same way that a command fails if an exclusive lock fails.
2015-07-17 23:13:22 +03:00
cmd - > lockd_vg_enforce_sh = 1 ;
}
2015-03-05 23:00:44 +03:00
2009-07-15 09:48:36 +04:00
return process_each_lv ( cmd , argc , argv ,
2010-06-28 23:10:16 +04:00
update ? READ_FOR_UPDATE : 0 , NULL ,
2014-04-01 22:08:08 +04:00
& _lvchange_single ) ;
2002-11-18 17:04:08 +03:00
}