Re: fstrim on raid1 LV with writemostly PV leads to system freeze

[Mike Snitzer <snitzer@kernel.org>]

[cc'ing Heinz and the linux-raid mailing list]

On Thu, Sep 21 2023 at  4:34P -0400,
Kirill Kirilenko <kirill@ultracoder.org> wrote:

> Hello.
> 
> I created two LVM physical volumes: one on NVMe device and one on SATA SSD.
> I added them to a volume group and created a logical RAID1 volume in it.
> Then I enabled 'writemostly' flag on the second (slowest) PV.
> And my system started to freeze at random times with no messages in syslog.
> I was able to determine that the freezing was happening during execution of
> 'fstrim' (via systemd timer). I checked this by running 'fstrim' manually.
> If I disable the 'writemostly' flag, I experience no freezes. I can
> reproduce this behavior on vanilla 6.5.0 kernel.
> 
> My LV is 150 GB ext4 volume, and it has lots of files in it, so running
> 'fstrim' takes around a minute. This may be important.
> 
> Additional information:
> OS: Linux Mint 21.2
> CPU: AMD Ryzen 7 5800X
> NVMe: Samsung SSD 980 500GB
> SATA SSD: Samsung SSD 850 EVO M.2 250GB
> 
> Best regards,
> Kirill Kirilenko.
> 

I just verified that 6.5.0 does have this DM core fix (needed to
prevent excessive splitting of discard IO.. which could cause fstrim
to take longer for a DM device), but again 6.5.0 has this fix so it
isn't relevant:
be04c14a1bd2 dm: use op specific max_sectors when splitting abnormal io

Given your use of 'writemostly' I'm inferring you're using lvm2's
raid1 that uses MD raid1 code in terms of the dm-raid target.

Discards (more generic term for fstrim) are considered writes, so
writemostly really shouldn't matter... but I know that there have been
issues with MD's writemostly code (identified by others relatively
recently).

All said: hopefully someone more MD oriented can review your report
and help you further.

Mike