On Sun, 22 Apr 2012 19:21:11 +0200 Jan Ceuleers <jan.ceuleers@xxxxxxxxxxxx> wrote: > NeilBrown wrote: > > Looks like: > > > > commit 7bfec5f35c68121e7b1849f3f4166dd96c8da5b3 > > > > is at fault. It causes md to attempt to add spares into the array more often. > > Would I be right in guessing that you have one spare in this array? > > If you remove the spare, the messages should stop. > > Hmmm. The commit message is as follows: > > commit 7bfec5f35c68121e7b1849f3f4166dd96c8da5b3 > Author: NeilBrown <neilb@xxxxxxx> > Date: Fri Dec 23 10:17:53 2011 +1100 > > md/raid5: If there is a spare and a want_replacement device, start > replaceme > > When attempting to add a spare to a RAID[456] array, also consider > adding it as a replacement for a want_replacement device. > > This requires that common md code attempt hot_add even when the array > is not formally degraded. > > Reviewed-by: Dan Williams <dan.j.williams@xxxxxxxxx> > Signed-off-by: NeilBrown <neilb@xxxxxxx> > > > Does this also apply to RAID1 (which is all I've got on this machine: no > RAID456)? Yes it does apply to RAID1. Part of the patch was RAID5-specific but part of it was to common code that would affect other levels. That part was not meant to be a big change, but it turned out to be a little bigger than I expected. The following should fix it. Thanks again for the report, NeilBrown From 321f820a905993f694f7ba4347492e9273831813 Mon Sep 17 00:00:00 2001 From: NeilBrown <neilb@xxxxxxx> Date: Mon, 23 Apr 2012 09:46:28 +1000 Subject: [PATCH] md: don't call ->add_disk unless there is good reason. Commit 7bfec5f35c68121e7b18 md/raid5: If there is a spare and a want_replacement device, start replacement. cause md_check_recovery to call ->add_disk much more often. Instead of only when the array is degraded, it is now called whenever md_check_recovery finds anything useful to do, which includes updating the metadata for clean<->dirty transition. This causes unnecessary work, and causes info messages from ->add_disk to be reported much too often. So refine md_check_recovery to only do any actual recovery checking (including ->add_disk) if MD_RECOVERY_NEEDED is set. This fix is suitable for 3.3.y: Cc: stable@xxxxxxxxxxxxxxx Reported-by: Jan Ceuleers <jan.ceuleers@xxxxxxxxxxxx> Signed-off-by: NeilBrown <neilb@xxxxxxx> diff --git a/drivers/md/md.c b/drivers/md/md.c index 9524192..47f1fdb6 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -7813,14 +7813,14 @@ void md_check_recovery(struct mddev *mddev) * any transients in the value of "sync_action". */ set_bit(MD_RECOVERY_RUNNING, &mddev->recovery); - clear_bit(MD_RECOVERY_NEEDED, &mddev->recovery); /* Clear some bits that don't mean anything, but * might be left set */ clear_bit(MD_RECOVERY_INTR, &mddev->recovery); clear_bit(MD_RECOVERY_DONE, &mddev->recovery); - if (test_bit(MD_RECOVERY_FROZEN, &mddev->recovery)) + if (!test_and_clear_bit(MD_RECOVERY_NEEDED, &mddev->recovery) || + test_bit(MD_RECOVERY_FROZEN, &mddev->recovery)) goto unlock; /* no recovery is running. * remove any failed drives, then
Attachment:
signature.asc
Description: PGP signature
