c/mirror_zfs
1
0
Fork 0
mirror of https://git.proxmox.com/git/mirror_zfs.git synced 2026-05-22 18:40:43 +03:00
Code Issues Packages Projects Releases Wiki Activity

Defer resilver only when progress is above a threshold

Browse Source 
Restart a resilver from scratch, if the current one in progress is
below a new tunable, zfs_resilver_defer_percent (defaulting to 10%).

The original rationale for deferring additional resilvers, when there is
already one in progress, was to help achieving data redundancy sooner
for the data that gets scanned at the end of the resilver.

But in case the admin wants to attach multiple disks to a single vdev,
it wasn't immediately obvious the admin is supposed to run
`zpool resilver` afterwards to reset the deferred resilvers and start
a new one from scratch.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Pavel Snajdr <snajpa@snajpa.net>
Closes #15810
This commit is contained in:
Pavel Snajdr
2024-10-04 19:41:17 +02:00
committed by GitHub
parent 3d0175d10e
commit 0d77e738e6
4 changed files with 62 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files
module/zfs/dsl_scan.c
+44 -19
View File
@@ -212,6 +212,9 @@ static uint64_t zfs_max_async_dedup_frees = 100000;
/* set to disable resilver deferring */
static int zfs_resilver_disable_defer = B_FALSE;
/* Don't defer a resilver if the one in progress only got this far: */
static uint_t zfs_resilver_defer_percent = 10;
/*
* We wait a few txgs after importing a pool to begin scanning so that
* the import / mounting code isn't held up by scrub / resilver IO.
@@ -4289,27 +4292,26 @@ dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx)
dsl_scan_t *scn = dp->dp_scan;
spa_t *spa = dp->dp_spa;
state_sync_type_t sync_type = SYNC_OPTIONAL;
int restart_early = 0;
if (spa->spa_resilver_deferred &&
!spa_feature_is_active(dp->dp_spa, SPA_FEATURE_RESILVER_DEFER))
spa_feature_incr(spa, SPA_FEATURE_RESILVER_DEFER, tx);
if (spa->spa_resilver_deferred) {
uint64_t to_issue, issued;
/*
* Check for scn_restart_txg before checking spa_load_state, so
* that we can restart an old-style scan while the pool is being
* imported (see dsl_scan_init). We also restart scans if there
* is a deferred resilver and the user has manually disabled
* deferred resilvers via the tunable.
*/
if (dsl_scan_restarting(scn, tx) ||
(spa->spa_resilver_deferred && zfs_resilver_disable_defer)) {
pool_scan_func_t func = POOL_SCAN_SCRUB;
dsl_scan_done(scn, B_FALSE, tx);
if (vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL))
func = POOL_SCAN_RESILVER;
zfs_dbgmsg("restarting scan func=%u on %s txg=%llu",
func, dp->dp_spa->spa_name, (longlong_t)tx->tx_txg);
dsl_scan_setup_sync(&func, tx);
if (!spa_feature_is_active(dp->dp_spa,
SPA_FEATURE_RESILVER_DEFER))
spa_feature_incr(spa, SPA_FEATURE_RESILVER_DEFER, tx);
/*
* See print_scan_scrub_resilver_status() issued/total_i
* @ cmd/zpool/zpool_main.c
*/
to_issue =
scn->scn_phys.scn_to_examine - scn->scn_phys.scn_skipped;
issued =
scn->scn_issued_before_pass + spa->spa_scan_pass_issued;
restart_early =
zfs_resilver_disable_defer ||
(issued < (to_issue * zfs_resilver_defer_percent / 100));
}
/*
@@ -4318,6 +4320,26 @@ dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx)
if (spa_sync_pass(spa) > 1)
return;
/*
* Check for scn_restart_txg before checking spa_load_state, so
* that we can restart an old-style scan while the pool is being
* imported (see dsl_scan_init). We also restart scans if there
* is a deferred resilver and the user has manually disabled
* deferred resilvers via zfs_resilver_disable_defer, or if the
* current scan progress is below zfs_resilver_defer_percent.
*/
if (dsl_scan_restarting(scn, tx) || restart_early) {
pool_scan_func_t func = POOL_SCAN_SCRUB;
dsl_scan_done(scn, B_FALSE, tx);
if (vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL))
func = POOL_SCAN_RESILVER;
zfs_dbgmsg("restarting scan func=%u on %s txg=%llu early=%d",
func, dp->dp_spa->spa_name, (longlong_t)tx->tx_txg,
restart_early);
dsl_scan_setup_sync(&func, tx);
}
/*
* If the spa is shutting down, then stop scanning. This will
* ensure that the scan does not dirty any new data during the
@@ -5287,6 +5309,9 @@ ZFS_MODULE_PARAM(zfs, zfs_, scan_report_txgs, UINT, ZMOD_RW,
ZFS_MODULE_PARAM(zfs, zfs_, resilver_disable_defer, INT, ZMOD_RW,
"Process all resilvers immediately");
ZFS_MODULE_PARAM(zfs, zfs_, resilver_defer_percent, UINT, ZMOD_RW,
"Issued IO percent complete after which resilvers are deferred");
ZFS_MODULE_PARAM(zfs, zfs_, scrub_error_blocks_per_txg, UINT, ZMOD_RW,
"Error blocks to be scrubbed in one txg");
/* END CSTYLED */