Linux-Raid Archives on lore.kernel.org
 help / color / Atom feed
From: Donald Buczek <buczek@molgen.mpg.de>
To: Guoqing Jiang <guoqing.jiang@cloud.ionos.com>,
	Song Liu <song@kernel.org>,
	linux-raid@vger.kernel.org,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	it+raid@molgen.mpg.de
Subject: Re: md_raid: mdX_raid6 looping after sync_action "check" to "idle" transition
Date: Tue, 1 Dec 2020 10:29:18 +0100
Message-ID: <77244d60-1c2d-330e-71e6-4907d4dd65fc@molgen.mpg.de> (raw)
In-Reply-To: <95fbd558-5e46-7a6a-43ac-bcc5ae8581db@cloud.ionos.com>

Am 30.11.20 um 03:06 schrieb Guoqing Jiang:
> 
> 
> On 11/28/20 13:25, Donald Buczek wrote:
>> Dear Linux mdraid people,
>>
>> we are using raid6 on several servers. Occasionally we had failures, where a mdX_raid6 process seems to go into a busy loop and all I/O to the md device blocks. We've seen this on various kernel versions.
>>
>> The last time this happened (in this case with Linux 5.10.0-rc4), I took some data.
>>
>> The triggering event seems to be the md_check cron job trying to pause the ongoing check operation in the morning with
>>
>>      echo idle > /sys/devices/virtual/block/md1/md/sync_action
>>
>> This doesn't complete. Here's /proc/stack of this process:
>>
>>      root@done:~/linux_problems/mdX_raid6_looping/2020-11-27# ps -fp 23333
>>      UID        PID  PPID  C STIME TTY          TIME CMD
>>      root     23333 23331  0 02:00 ?        00:00:00 /bin/bash /usr/bin/mdcheck --continue --duration 06:00
>>      root@done:~/linux_problems/mdX_raid6_looping/2020-11-27# cat /proc/23333/stack
>>      [<0>] kthread_stop+0x6e/0x150
>>      [<0>] md_unregister_thread+0x3e/0x70
>>      [<0>] md_reap_sync_thread+0x1f/0x1e0
>>      [<0>] action_store+0x141/0x2b0
>>      [<0>] md_attr_store+0x71/0xb0
>>      [<0>] kernfs_fop_write+0x113/0x1a0
>>      [<0>] vfs_write+0xbc/0x250
>>      [<0>] ksys_write+0xa1/0xe0
>>      [<0>] do_syscall_64+0x33/0x40
>>      [<0>] entry_SYSCALL_64_after_hwframe+0x44/0xa9
>>
>> Note, that md0 has been paused successfully just before.
> 
> What is the personality of md0? Is it also raid6?

Yes.

> 
>>
>>      2020-11-27T02:00:01+01:00 done CROND[23333]: (root) CMD (/usr/bin/mdcheck --continue --duration "06:00")
>>      2020-11-27T02:00:01+01:00 done root: mdcheck continue checking /dev/md0 from 10623180920
>>      2020-11-27T02:00:01.382994+01:00 done kernel: [378596.606381] md: data-check of RAID array md0
>>      2020-11-27T02:00:01+01:00 done root: mdcheck continue checking /dev/md1 from 11582849320
>>      2020-11-27T02:00:01.437999+01:00 done kernel: [378596.661559] md: data-check of RAID array md1
>>      2020-11-27T06:00:01.842003+01:00 done kernel: [392996.625147] md: md0: data-check interrupted.
>>      2020-11-27T06:00:02+01:00 done root: pause checking /dev/md0 at 13351127680
>>      2020-11-27T06:00:02.338989+01:00 done kernel: [392997.122520] md: md1: data-check interrupted.
>>      [ nothing related following.... ]
>>
>> After that, we see md1_raid6 in a busy loop:
>>
>>      PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
>>      2376 root     20   0       0      0      0 R 100.0  0.0   1387:38 md1_raid6
> 
> Seems the reap sync thread was trying to stop md1_raid6 while md1_raid6 was triggered again and again.
> 
>>
>> Also, all processes doing I/O do the md device block.
>>
>> This is /proc/mdstat:
>>
>>      Personalities : [linear] [raid0] [raid1] [raid6] [raid5] [raid4] [multipath]
>>      md1 : active raid6 sdk[0] sdj[15] sdi[14] sdh[13] sdg[12] sdf[11] sde[10] sdd[9] sdc[8] sdr[7] sdq[6] sdp[5] sdo[4] sdn[3] sdm[2] sdl[1]
>>            109394518016 blocks super 1.2 level 6, 512k chunk, algorithm 2 [16/16] [UUUUUUUUUUUUUUUU]
>>            [==================>..]  check = 94.0% (7350290348/7813894144) finish=57189.3min speed=135K/sec
>>            bitmap: 0/59 pages [0KB], 65536KB chunk
>>      md0 : active raid6 sds[0] sdah[15] sdag[16] sdaf[13] sdae[12] sdad[11] sdac[10] sdab[9] sdaa[8] sdz[7] sdy[6] sdx[17] sdw[4] sdv[3] sdu[2] sdt[1]
>>            109394518016 blocks super 1.2 level 6, 512k chunk, algorithm 2 [16/16] [UUUUUUUUUUUUUUUU]
>>            bitmap: 0/59 pages [0KB], 65536KB chunk
>>
> 
> So the RECOVERY_CHECK flag should be set, not sure if the simple changes
> helps, but you may give it a try.

Thanks. I've copied the original condition block to execute before the modified one and added some logging to it to see, if the change actually triggers. I will pause and unpause the check frequently on a busy machine to get this code executed more often.

Best
   Donald

> 
> diff --git a/drivers/md/md.c b/drivers/md/md.c
> index 98bac4f..e2697d0 100644
> --- a/drivers/md/md.c
> +++ b/drivers/md/md.c
> @@ -9300,7 +9300,8 @@ void md_check_recovery(struct mddev *mddev)
>                          md_update_sb(mddev, 0);
> 
>                  if (test_bit(MD_RECOVERY_RUNNING, &mddev->recovery) &&
> -                   !test_bit(MD_RECOVERY_DONE, &mddev->recovery)) {
> +                   (!test_bit(MD_RECOVERY_DONE, &mddev->recovery) ||
> +                    test_bit(MD_RECOVERY_CHECK, &mddev->recovery))) {
>                          /* resync/recovery still happening */
>                          clear_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
>                          goto unlock;
> 
> Thanks,
> Guoqing

-- 
Donald Buczek
buczek@molgen.mpg.de
Tel: +49 30 8413 1433

  reply index

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-11-28 12:25 Donald Buczek
2020-11-30  2:06 ` Guoqing Jiang
2020-12-01  9:29   ` Donald Buczek [this message]
2020-12-02 17:28     ` Donald Buczek
2020-12-03  1:55       ` Guoqing Jiang
2020-12-03 11:42         ` Donald Buczek
2020-12-21 12:33           ` Donald Buczek
2021-01-19 11:30             ` Donald Buczek
2021-01-20 16:33               ` Guoqing Jiang
2021-01-23 13:04                 ` Donald Buczek
2021-01-25  8:54                   ` Donald Buczek
2021-01-25 21:32                     ` Donald Buczek
2021-01-26  0:44                       ` Guoqing Jiang
2021-01-26  9:50                         ` Donald Buczek
2021-01-26 11:14                           ` Guoqing Jiang
2021-01-26 12:58                             ` Donald Buczek
2021-01-26 14:06                               ` Guoqing Jiang
2021-01-26 16:05                                 ` Donald Buczek
2021-02-02 15:42                                   ` Guoqing Jiang
2021-02-08 11:38                                     ` Donald Buczek
2021-02-08 14:53                                       ` Guoqing Jiang
2021-02-08 18:41                                         ` Donald Buczek
2021-02-09  0:46                                           ` Guoqing Jiang
2021-02-09  9:24                                             ` Donald Buczek

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=77244d60-1c2d-330e-71e6-4907d4dd65fc@molgen.mpg.de \
    --to=buczek@molgen.mpg.de \
    --cc=guoqing.jiang@cloud.ionos.com \
    --cc=it+raid@molgen.mpg.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-raid@vger.kernel.org \
    --cc=song@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Linux-Raid Archives on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-raid/0 linux-raid/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-raid linux-raid/ https://lore.kernel.org/linux-raid \
		linux-raid@vger.kernel.org
	public-inbox-index linux-raid

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-raid


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git