[LU-15244] sanity test_101j: FAIL: expected 4096 got 4100

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.15.0
• Fix Version/s: Lustre 2.15.0
• Type: Bug
• Priority: Minor
• Reporter: Maloo
• Assignee: Qian Yingjin
• Resolution: Fixed

Environment

RHEL 8.5

• Severity: 3

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/ecec4ec7-16ee-4a58-a971-711126b77faf

test_101j failed with the following error:

== sanity test 101j: A complete read block should be submitted when no RA ========================================================== 19:56:32 (1637092592)
Disable read-ahead
16+0 records in
16+0 records out
16777216 bytes (17 MB, 16 MiB) copied, 0.0582022 s, 288 MB/s
Reset readahead stats
4096+0 records in
4096+0 records out
16777216 bytes (17 MB, 16 MiB) copied, 1.02446 s, 16.4 MB/s
snapshot_time             3997.129812966 secs.nsecs
start_time                0.000000000 secs.nsecs
elapsed_time              3997.129812966 secs.nsecs
failed_to_fast_read       4100 samples [pages]
 sanity test_101j: @@@@@@ FAIL: expected 4096 got 4100

VVVVVVV DO NOT REMOVE LINES BELOW, Added by Maloo for auto-association VVVVVVV
sanity test_101j - expected 4096 got 4100

Comments

[Jian Yu]

This is a regression failure on RHEL 8.5 distro.
sanity test 101j was introduced by the following patch: https://review.whamcloud.com/35217 "LU-12043 llite: fix to submit complete read block with ra disabled".

[Qian Yingjin]

Run sanity.sh test_101j locally, can not reproduce the bug...

[Jian Yu]

Hi Yingjin,
Not sure if it's caused by previous sanity sub-tests. The failure can be reproduced in autotest runs. You might submit a debug patch to gather debug info from autotest run.

[Qian Yingjin]

Hi Jian,

I search failure ratio in Maloo for sanity 101j, it is 0 for the recent 100 runs.
Does this failure only happen on rhel8.5, or other versions of RHEL also can trigger this failure?

Regards,
Qian

[Peter Jones]

Yingjin

Jian said

"This is a regression failure on RHEL 8.5 distro.
sanity test 101j was introduced by the following patch: https://review.whamcloud.com/35217 "LU-12043 llite: fix to submit complete read block with ra disabled"."

To me this sounds like a regression for RHEL 8.5. Given that RHEL 8.5 is only now being introduced that would explain the low incidence in recent runs. Did you use RHEL 8.5 locally when you tried to reproduce?

Peter

[Qian Yingjin]

No, I used rhel8.1. Let me try rhel 8.5 locally.

[Qian Yingjin]

I reproduced this bug locally on CentOS 8.5.

Will investigate the reason.

[Gerrit Updater]

"Yingjin Qian <[e-mail]>" uploaded a new patch: https://review.whamcloud.com/45712
Subject: LU-15244 llite: set ra_pages of backing_dev_info with 0
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: c1527f2b6b1cbe5b478630a967eacc4c316f24cd

[Gerrit Updater]

"Oleg Drokin <[e-mail]>" merged in patch https://review.whamcloud.com/45712/
Subject: LU-15244 llite: set ra_pages of backing_dev_info with 0
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 878561880d2aba038db95e199f82b186f22daa45

[Peter Jones]

Landed for 2.15

[Gerrit Updater]

"Jian Yu <[e-mail]>" uploaded a new patch: https://review.whamcloud.com/45969
Subject: LU-15244 llite: set ra_pages of backing_dev_info with 0
Project: fs/lustre-release
Branch: b2_14
Current Patch Set: 1
Commit: 6a4e98a11884868e8d4368ce5540454378c42ef9

[Gerrit Updater]

"Jian Yu <[e-mail]>" uploaded a new patch: https://review.whamcloud.com/45970
Subject: LU-15244 llite: set ra_pages of backing_dev_info with 0
Project: fs/lustre-release
Branch: b2_12
Current Patch Set: 1
Commit: 066214fa758314a45b8af28ee076ad3607c83b9e

[Gerrit Updater]

"Oleg Drokin <[e-mail]>" merged in patch https://review.whamcloud.com/45969/
Subject: LU-15244 llite: set ra_pages of backing_dev_info with 0
Project: fs/lustre-release
Branch: b2_14
Current Patch Set:
Commit: 7111122431ff6ea7edcf073afd63a13293bc7a51