[dm-devel] [PATCH RFC] dm-raid1: keep writing after leg failure

[Lidong Zhong]

Currently if there is a leg failure, the bio will be put into the hold
list until userspace replace/remove the leg. Here we are trying to make
dm-raid1 ignore the failure and keep the following bios going on.
This is because there maybe a temporary path failure in clvmd
which leads to cluster raid1 remove/replace the fake device failure. And
it takes a long time to do the full sync if we readd the device back.
---
 drivers/md/dm-raid1.c | 30 ++++++++++++++++++++++++++----
 1 file changed, 26 insertions(+), 4 deletions(-)

diff --git a/drivers/md/dm-raid1.c b/drivers/md/dm-raid1.c
index 9584443..e237c42 100644
--- a/drivers/md/dm-raid1.c
+++ b/drivers/md/dm-raid1.c
@@ -24,7 +24,9 @@
 #define MAX_RECOVERY 1	/* Maximum number of regions recovered in parallel. */
 
 #define DM_RAID1_HANDLE_ERRORS 0x01
+#define DM_RAID1_KEEP_LOG      0x02
 #define errors_handled(p)	((p)->features & DM_RAID1_HANDLE_ERRORS)
+#define keep_log(p)	        ((p)->features & DM_RAID1_KEEP_LOG)
 
 static DECLARE_WAIT_QUEUE_HEAD(_kmirrord_recovery_stopped);
 
@@ -750,7 +752,7 @@ static void do_writes(struct mirror_set *ms, struct bio_list *writes)
 		dm_rh_delay(ms->rh, bio);
 
 	while ((bio = bio_list_pop(&nosync))) {
-		if (unlikely(ms->leg_failure) && errors_handled(ms)) {
+		if (unlikely(ms->leg_failure) && errors_handled(ms) && !keep_log(ms)) {
 			spin_lock_irq(&ms->lock);
 			bio_list_add(&ms->failures, bio);
 			spin_unlock_irq(&ms->lock);
@@ -800,9 +802,19 @@ static void do_failures(struct mirror_set *ms, struct bio_list *failures)
 		 * be wrong if the failed leg returned after reboot and
 		 * got replicated back to the good legs.)
 		 */
-		if (!get_valid_mirror(ms))
+
+		/*
+		 * we return EIO when the log device is failed if keep_log is set
+		 */
+		if (!get_valid_mirror(ms) || (keep_log(ms) && !ms->log_failure))
 			bio_endio(bio, -EIO);
-		else if (errors_handled(ms))
+		/*
+		 * After the userspace get noticed that the leg has failed,
+		 * we just pretend that the bio has suceeded since the region 
+		 * has already been marked nosync. It's OK do the recovery after
+		 * the device comes back
+		 */
+		else if (errors_handled(ms) && !keep_log(ms))
 			hold_bio(ms, bio);
 		else
 			bio_endio(bio, 0);
@@ -1005,8 +1017,15 @@ static int parse_features(struct mirror_set *ms, unsigned argc, char **argv,
 		return -EINVAL;
 	}
 
+	argc--;
+	argv++;
 	(*args_used)++;
 
+	if (!strcmp("keep_log", argv[0])) {
+		ms->features |= DM_RAID1_KEEP_LOG;
+		(*args_used)++;
+	}
+
 	return 0;
 }
 
@@ -1382,8 +1401,11 @@ static void mirror_status(struct dm_target *ti, status_type_t type,
 			DMEMIT(" %s %llu", ms->mirror[m].dev->name,
 			       (unsigned long long)ms->mirror[m].offset);
 
-		if (ms->features & DM_RAID1_HANDLE_ERRORS)
+		if (errors_handled(ms) && keep_log(ms))
+			DMEMIT(" 2 handle_errors keep_log");
+		else if (errors_handled(ms))
 			DMEMIT(" 1 handle_errors");
+
 	}
 }
 
-- 
1.8.1.4