[LU-10155] File creation with 'lfs setstripe' is not replayable

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.11.0
• Fix Version/s: Lustre 2.11.0
• Type: Bug
• Priority: Blocker
• Reporter: Mikhail Pershin
• Assignee: Lai Siyao
• Resolution: Fixed
• Severity: 3

Description

File created with 'lfs setstripe' cannot recover from failover. Simple test like the one below fails:

test_00() {
replay_barrier $SINGLEMDS
$LFS setstripe -c 2 $DIR/$tfile
fail $SINGLEMDS
$CHECKSTAT -t file $DIR/$tfile ||
		error "$CHECKSTAT $DIR/$tfile check failed"
}

At first sign the file is not re-created due to problem with LOV EA supplied along with replay, it says LOV MAGIC is 0.

Issue Links

• is related to: LU-9514 sanity test 51f times out due to out ...

Comments

[Andreas Dilger]

Mike, do you know when this problem was introduced? How is it possible that this wasn't detected by our testing?

[Joseph Gmitter (Inactive)]

Hi Lai,

Can you please investigate?

Thanks.
Joe

[Jinshan Xiong (Inactive)]

A recent change to this area of code is LU-9514, which tried to optimize memory usage by freeing the open reply buffer earlier.

[Mikhail Pershin]

I have tried to revert that patch but it doesn't help

[Lai Siyao]

I can't reproduce in latest master, maybe this was fixed in other tickets, but anyway I'll add this test into replay-single.sh.

[Gerrit Updater]

Lai Siyao ([e-mail]) uploaded a new patch: https://review.whamcloud.com/30704
Subject: LU-10155 test: add setstripe replay test
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 8341c5f2499aa22bd230c9c082cc7dc82b821bf7

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/30704/
Subject: LU-10155 recovery: support setstripe replay
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 3a0a50f44f68066642a23365fa4a6f1924dd1108

[Peter Jones]

Mark as resolved for now I suppose - we will reopen if the test starts failing now it is landed