|
 |
5eacff |
From a4e96fd8f3f0b5416783237c1cb6ee87e7eff23d Mon Sep 17 00:00:00 2001
|
|
|
5eacff |
From: Mariusz Tkaczyk <mariusz.tkaczyk@intel.com>
|
|
|
5eacff |
Date: Fri, 8 Feb 2019 11:07:10 +0100
|
|
|
5eacff |
Subject: [RHEL7.7 PATCH 12/24] imsm: finish recovery when drive with rebuild
|
|
|
5eacff |
fails
|
|
|
5eacff |
|
|
|
5eacff |
Commit d7a1fda2769b ("imsm: update metadata correctly while raid10 double
|
|
|
5eacff |
degradation") resolves main Imsm double degradation problems but it
|
|
|
5eacff |
omits one case. Now metadata hangs in the rebuilding state if the drive
|
|
|
5eacff |
under rebuild is removed during recovery from double degradation.
|
|
|
5eacff |
|
|
|
5eacff |
The root cause of this problem is comparing new map_state with current
|
|
|
5eacff |
and if they both are degraded assuming that nothing new happens.
|
|
|
5eacff |
|
|
|
5eacff |
Don't rely on map states, just check if device is failed. If the drive
|
|
|
5eacff |
under rebuild fails then finish migration, in other cases update map
|
|
|
5eacff |
state only (second fail means that destination map state can't be normal).
|
|
|
5eacff |
|
|
|
5eacff |
To avoid problems with reassembling move end_migration (called after
|
|
|
5eacff |
double degradation successful recovery) after check if recovery really
|
|
|
5eacff |
finished, for details see (7ce057018 "imsm: fix: rebuild does not
|
|
|
5eacff |
continue after reboot").
|
|
|
5eacff |
Remove redundant code responsible for finishing rebuild process. Function
|
|
|
5eacff |
end_migration do exactly the same. Set last_checkpoint to 0, to prepare
|
|
|
5eacff |
it for the next rebuild.
|
|
|
5eacff |
|
|
|
5eacff |
Signed-off-by: Mariusz Tkaczyk <mariusz.tkaczyk@intel.com>
|
|
|
5eacff |
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
|
|
|
5eacff |
---
|
|
|
5eacff |
super-intel.c | 26 +++++++++++---------------
|
|
|
5eacff |
1 file changed, 11 insertions(+), 15 deletions(-)
|
|
|
5eacff |
|
|
|
5eacff |
diff --git a/super-intel.c b/super-intel.c
|
|
|
5eacff |
index d2035cc..38a1b6c 100644
|
|
|
5eacff |
--- a/super-intel.c
|
|
|
5eacff |
+++ b/super-intel.c
|
|
|
5eacff |
@@ -8560,26 +8560,22 @@ static void imsm_set_disk(struct active_array *a, int n, int state)
|
|
|
5eacff |
}
|
|
|
5eacff |
if (is_rebuilding(dev)) {
|
|
|
5eacff |
dprintf_cont("while rebuilding ");
|
|
|
5eacff |
- if (map->map_state != map_state) {
|
|
|
5eacff |
- dprintf_cont("map state change ");
|
|
|
5eacff |
+ if (state & DS_FAULTY) {
|
|
|
5eacff |
+ dprintf_cont("removing failed drive ");
|
|
|
5eacff |
if (n == map->failed_disk_num) {
|
|
|
5eacff |
dprintf_cont("end migration");
|
|
|
5eacff |
end_migration(dev, super, map_state);
|
|
|
5eacff |
+ a->last_checkpoint = 0;
|
|
|
5eacff |
} else {
|
|
|
5eacff |
- dprintf_cont("raid10 double degradation, map state change");
|
|
|
5eacff |
+ dprintf_cont("fail detected during rebuild, changing map state");
|
|
|
5eacff |
map->map_state = map_state;
|
|
|
5eacff |
}
|
|
|
5eacff |
super->updates_pending++;
|
|
|
5eacff |
- } else if (!rebuild_done)
|
|
|
5eacff |
- break;
|
|
|
5eacff |
- else if (n == map->failed_disk_num) {
|
|
|
5eacff |
- /* r10 double degraded to degraded transition */
|
|
|
5eacff |
- dprintf_cont("raid10 double degradation end migration");
|
|
|
5eacff |
- end_migration(dev, super, map_state);
|
|
|
5eacff |
- a->last_checkpoint = 0;
|
|
|
5eacff |
- super->updates_pending++;
|
|
|
5eacff |
}
|
|
|
5eacff |
|
|
|
5eacff |
+ if (!rebuild_done)
|
|
|
5eacff |
+ break;
|
|
|
5eacff |
+
|
|
|
5eacff |
/* check if recovery is really finished */
|
|
|
5eacff |
for (mdi = a->info.devs; mdi ; mdi = mdi->next)
|
|
|
5eacff |
if (mdi->recovery_start != MaxSector) {
|
|
|
5eacff |
@@ -8588,7 +8584,7 @@ static void imsm_set_disk(struct active_array *a, int n, int state)
|
|
|
5eacff |
}
|
|
|
5eacff |
if (recovery_not_finished) {
|
|
|
5eacff |
dprintf_cont("\n");
|
|
|
5eacff |
- dprintf_cont("Rebuild has not finished yet, map state changes only if raid10 double degradation happens");
|
|
|
5eacff |
+ dprintf_cont("Rebuild has not finished yet");
|
|
|
5eacff |
if (a->last_checkpoint < mdi->recovery_start) {
|
|
|
5eacff |
a->last_checkpoint =
|
|
|
5eacff |
mdi->recovery_start;
|
|
|
5eacff |
@@ -8598,9 +8594,9 @@ static void imsm_set_disk(struct active_array *a, int n, int state)
|
|
|
5eacff |
}
|
|
|
5eacff |
|
|
|
5eacff |
dprintf_cont(" Rebuild done, still degraded");
|
|
|
5eacff |
- dev->vol.migr_state = 0;
|
|
|
5eacff |
- set_migr_type(dev, 0);
|
|
|
5eacff |
- dev->vol.curr_migr_unit = 0;
|
|
|
5eacff |
+ end_migration(dev, super, map_state);
|
|
|
5eacff |
+ a->last_checkpoint = 0;
|
|
|
5eacff |
+ super->updates_pending++;
|
|
|
5eacff |
|
|
|
5eacff |
for (i = 0; i < map->num_members; i++) {
|
|
|
5eacff |
int idx = get_imsm_ord_tbl_ent(dev, i, MAP_0);
|
|
|
5eacff |
--
|
|
|
5eacff |
2.7.5
|
|
|
5eacff |
|