Commit dafdb289 authored by Artur Paszkiewicz's avatar Artur Paszkiewicz Committed by Jim Harris
Browse files

raid: allow re-adding a base bdev with superblock 



If a raid bdev is in ONLINE state and a base bdev is added that was a
member of this array, rebuild should start automatically.

Signed-off-by: default avatarArtur Paszkiewicz <artur.paszkiewicz@intel.com>
Change-Id: I30583ced59f351658d1c5029faea49fbb2ee9367
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/22501


Reviewed-by: default avatarKonrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: default avatarJim Harris <jim.harris@samsung.com>
Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
parent b694ff86

module/bdev/raid/bdev_raid.c

+30 −1
Original line number Diff line number Diff line
@@ -3008,6 +3008,9 @@ raid_bdev_configure_base_bdev_cont(struct raid_base_bdev_info *base_info) 
	}

}



static void raid_bdev_examine_sb(const struct raid_bdev_superblock *sb, struct spdk_bdev *bdev,

				 raid_base_bdev_cb cb_fn, void *cb_ctx);



static void

raid_bdev_configure_base_bdev_check_sb_cb(const struct raid_bdev_superblock *sb, int status,

		void *ctx)
@@ -3017,7 +3020,14 @@ raid_bdev_configure_base_bdev_check_sb_cb(const struct raid_bdev_superblock *sb, 
	switch (status) {

	case 0:

		/* valid superblock found */

		SPDK_ERRLOG("Existing raid superblock found on bdev %s\n", base_info->name);

		if (spdk_uuid_compare(&base_info->raid_bdev->bdev.uuid, &sb->uuid) == 0) {

			struct spdk_bdev *bdev = spdk_bdev_desc_get_bdev(base_info->desc);



			raid_bdev_free_base_bdev_resource(base_info);

			raid_bdev_examine_sb(sb, bdev, base_info->configure_cb, base_info->configure_cb_ctx);

			return;

		}

		SPDK_ERRLOG("Superblock of a different raid bdev found on bdev %s\n", base_info->name);

		status = -EEXIST;

		raid_bdev_free_base_bdev_resource(base_info);

		break;
@@ -3431,6 +3441,25 @@ raid_bdev_examine_sb(const struct raid_bdev_superblock *sb, struct spdk_bdev *bd 
		}

	}



	if (raid_bdev->state == RAID_BDEV_STATE_ONLINE) {

		assert(sb_base_bdev->slot < raid_bdev->num_base_bdevs);

		base_info = &raid_bdev->base_bdev_info[sb_base_bdev->slot];

		assert(base_info->desc == NULL);

		assert(sb_base_bdev->state == RAID_SB_BASE_BDEV_MISSING ||

		       sb_base_bdev->state == RAID_SB_BASE_BDEV_FAILED);

		assert(spdk_uuid_is_null(&base_info->uuid));

		spdk_uuid_copy(&base_info->uuid, &sb_base_bdev->uuid);

		SPDK_NOTICELOG("Re-adding bdev %s to raid bdev %s.\n", bdev->name, raid_bdev->bdev.name);

		spdk_spin_lock(&raid_bdev->base_bdev_lock);

		rc = raid_bdev_configure_base_bdev(base_info, true, cb_fn, cb_ctx);

		spdk_spin_unlock(&raid_bdev->base_bdev_lock);

		if (rc != 0) {

			SPDK_ERRLOG("Failed to configure bdev %s as base bdev of raid %s: %s\n",

				    bdev->name, raid_bdev->bdev.name, spdk_strerror(-rc));

		}

		goto out;

	}



	if (sb_base_bdev->state != RAID_SB_BASE_BDEV_CONFIGURED) {

		SPDK_NOTICELOG("Bdev %s is not an active member of raid bdev %s. Ignoring.\n",

			       bdev->name, raid_bdev->bdev.name);

test/bdev/bdev_raid.sh

+30 −0
Original line number Diff line number Diff line
@@ -760,6 +760,36 @@ function raid_rebuild_test() { 
		verify_raid_bdev_state $raid_bdev_name "online" $raid_level $strip_size $num_base_bdevs_operational

		verify_raid_bdev_process $raid_bdev_name "none" "none"

		[[ $($rpc_py bdev_raid_get_bdevs all | jq -r '.[].base_bdevs_list[0].name') == "spare" ]]



		# Remove and re-add a base bdev - rebuild should start automatically

		$rpc_py bdev_raid_remove_base_bdev "spare"

		verify_raid_bdev_state $raid_bdev_name "online" $raid_level $strip_size $((num_base_bdevs_operational - 1))

		$rpc_py bdev_raid_add_base_bdev $raid_bdev_name "spare"

		sleep 1

		verify_raid_bdev_process $raid_bdev_name "rebuild" "spare"



		# Same as above but re-add through examine

		$rpc_py bdev_passthru_delete "spare"

		verify_raid_bdev_state $raid_bdev_name "online" $raid_level $strip_size $((num_base_bdevs_operational - 1))

		$rpc_py bdev_passthru_create -b "spare_delay" -p "spare"

		sleep 1

		verify_raid_bdev_process $raid_bdev_name "rebuild" "spare"



		# Stop the rebuild

		$rpc_py bdev_passthru_delete "spare"

		verify_raid_bdev_state $raid_bdev_name "online" $raid_level $strip_size $((num_base_bdevs_operational - 1))

		verify_raid_bdev_process $raid_bdev_name "none" "none"



		# Re-adding a base bdev that was replaced (no longer is a member of the array) should not be allowed

		$rpc_py bdev_passthru_delete ${base_bdevs[0]}

		$rpc_py bdev_passthru_create -b ${base_bdevs[0]}_malloc -p ${base_bdevs[0]}

		sleep 1

		verify_raid_bdev_state $raid_bdev_name "online" $raid_level $strip_size $((num_base_bdevs_operational - 1))

		verify_raid_bdev_process $raid_bdev_name "none" "none"

		NOT $rpc_py bdev_raid_add_base_bdev $raid_bdev_name ${base_bdevs[0]}

		sleep 1

		verify_raid_bdev_state $raid_bdev_name "online" $raid_level $strip_size $((num_base_bdevs_operational - 1))

		verify_raid_bdev_process $raid_bdev_name "none" "none"

	fi



	killprocess $raid_pid