Merge 26818ada36 into b3b7491615

man/man4/zfs.4

@@ -1961,6 +1961,13 @@ Ignore the
 feature, causing an operation that would start a resilver to
 immediately restart the one in progress.
 .
+.It Sy zfs_resilver_defer_percent Ns = Ns Sy 10 Ns % Pq uint
+If the ongoing resilver progress is below this threshold, a new resilver will
+restart from scratch instead of being deferred after the current one finishes,
+even if the
+.Sy resilver_defer
+feature is enabled.
+.
 .It Sy zfs_resilver_min_time_ms Ns = Ns Sy 3000 Ns ms Po 3 s Pc Pq uint
 Resilvers are processed by the sync thread.
 While resilvering, it will spend at least this much time

module/zfs/dsl_scan.c

@@ -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.
@@ -4287,35 +4290,49 @@ 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;
+	uint64_t to_issue, issued;
+	int restart_early;
 
 	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);
 
-	/*
-	 * 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);
-	}
-
 	/*
 	 * Only process scans in sync pass 1.
 	 */
 	if (spa_sync_pass(spa) > 1)
 		return;
 
+	/*
+	 * issued/to_issue as presented to the user
+	 * in 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 = spa->spa_resilver_deferred && (
+	    zfs_resilver_disable_defer ||
+	    (issued < (to_issue * zfs_resilver_defer_percent / 100)));
+
+	/*
+	 * 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
+	 * curent 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
@@ -5285,6 +5302,9 @@ 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 */

tests/zfs-tests/include/tunables.cfg

@@ -73,6 +73,7 @@ REBUILD_SCRUB_ENABLED		rebuild_scrub_enabled		zfs_rebuild_scrub_enabled
 REMOVAL_SUSPEND_PROGRESS	removal_suspend_progress	zfs_removal_suspend_progress
 REMOVE_MAX_SEGMENT		remove_max_segment		zfs_remove_max_segment
 RESILVER_MIN_TIME_MS		resilver_min_time_ms		zfs_resilver_min_time_ms
+RESILVER_DEFER_PERCENT		resilver_defer_percent		zfs_resilver_defer_percent
 SCAN_LEGACY			scan_legacy			zfs_scan_legacy
 SCAN_SUSPEND_PROGRESS		scan_suspend_progress		zfs_scan_suspend_progress
 SCAN_VDEV_LIMIT			scan_vdev_limit			zfs_scan_vdev_limit

tests/zfs-tests/tests/functional/replacement/resilver_restart_001.ksh

@@ -96,6 +96,8 @@ set -A RESTARTS -- '1' '2' '2' '2'
 set -A VDEVS -- '' '' '' ''
 set -A DEFER_RESTARTS -- '1' '1' '1' '2'
 set -A DEFER_VDEVS -- '-' '2' '2' '-'
+set -A EARLY_RESTART_DEFER_RESTARTS -- '1' '2' '2' '2'
+set -A EARLY_RESTART_DEFER_VDEVS -- '' '' '' ''
 
 VDEV_REPLACE="${VDEV_FILES[1]} $SPARE_VDEV_FILE"
 
@@ -125,7 +127,7 @@ done
 wait
 
 # test without and with deferred resilve feature enabled
-for test in "without" "with"
+for test in "without" "with" "with_early_restart"
 do
 	log_note "Testing $test deferred resilvers"
 
@@ -135,6 +137,13 @@ do
 		RESTARTS=( "${DEFER_RESTARTS[@]}" )
 		VDEVS=( "${DEFER_VDEVS[@]}" )
 		VDEV_REPLACE="$SPARE_VDEV_FILE ${VDEV_FILES[1]}"
+		log_must set_tunable32 RESILVER_DEFER_PERCENT 0
+	elif [[ $test == "with_early_restart" ]]
+	then
+		RESTARTS=( "${EARLY_RESTART_DEFER_RESTARTS[@]}" )
+		VDEVS=( "${EARLY_RESTART_DEFER_VDEVS[@]}" )
+		VDEV_REPLACE="${VDEV_FILES[1]} $SPARE_VDEV_FILE"
+		log_must set_tunable32 RESILVER_DEFER_PERCENT 100
 	fi
 
 	# clear the events