[LU-17927] sanity test_119e: FAIL: trivial unaligned dio failed

• Type: Bug
• Resolution: Cannot Reproduce
• Priority: Critical
• Affects Version/s: Lustre 2.16.0
• Environment: RHEL 9.4 client
• Severity: 3
• Assignee: Shaun Tancheff
• Reporter: Maloo

Description

This issue was created by maloo for jianyu <[e-mail]>

This issue relates to the following test suite run: https://testing.whamcloud.com/test_sets/36e444c9-8e5c-48de-a4a2-bb198ffc9fe6

test_119e failed with the following error:

== sanity test 119e: Basic tests of dio read and write at various sizes ========================================================== 23:19:26 (1718061566)
1+0 records in
1+0 records out
26214400 bytes (26 MB, 25 MiB) copied, 0.681171 s, 38.5 MB/s
dd: error reading '/mnt/lustre/f119e.sanity.1': Invalid argument
0+0 records in
0+0 records out
0 bytes copied, 0.00488452 s, 0.0 kB/s
 sanity test_119e: @@@@@@ FAIL: trivial unaligned dio failed

Test session details:
clients: https://build.whamcloud.com/job/lustre-reviews/105356 - 5.14.0-427.16.1.el9_4.x86_64
servers: https://build.whamcloud.com/job/lustre-reviews/105356 - 5.14.0-362.24.1_lustre.el9.x86_64

<<Please provide additional information about the failure here>>

VVVVVVV DO NOT REMOVE LINES BELOW, Added by Maloo for auto-association VVVVVVV
sanity test_119e - trivial unaligned dio failed

Issue Links

• is blocked by: LU-17085 llite: safely duplicate iov_iter (was: pin_user_pages on unaligned DIO/AIO)
• is related to: LU-17168 sanity test_119f: dio copy failed
• is related to: LU-17289 el9.3 sanity test_906: fio seqwrite /mnt/lustre/f906.sanity failed
• is related to: LU-17525 Unaligned DIO interop with different page sizes fails
• is related to: LU-17404 Support RHEL 9.4

Comments

[Peter Jones]

According to a comment from Patrick in the above changeset it was rolled into the RHEL 9.4 server support patch which has been merged. So is there anything still remaining to be tracked by this ticket? 

[Gerrit Updater]

"Patrick Farrell <[e-mail]>" uploaded a new patch: https://review.whamcloud.com/c/fs/lustre-release/+/55650
Subject: LU-17927 tests: remove dio always_except for el9.4
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 1c61d064db37b31bb0e1f7409435efc885b7f992

[Andreas Dilger]

Depending on which patch is landing first, the always_except for test_119 should be removed either from https://review.whamcloud.com/54712 or by one of the UDIO patches.

[Shaun Tancheff]

fwiw:
https://review.whamcloud.com/c/fs/lustre-release/+/53997
and
https://review.whamcloud.com/c/fs/lustre-release/+/52266
are probably both needed for el9.4, the later needed an el9.4 specific fix.

Without the above patches el9.4 crashes, with the patches all the 119 series tests pass.

[Andreas Dilger]

paf, stancheff, this seems like a blocker for UDIO in 2.16.0 if this is failing consistently with el9.4.

Is this addressed by Shaun's in-flight UDIO fix patch in LU-17525, or is there something new in the el9.4 kernel that is causing problems for UDIO?

[Jian Yu]

sanity test 119f, 119g, and 119h hit the similar failure.