[LU-11671] sanity test 45: FAIL: write wasn't cached

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.16.0
• Affects Version/s: Lustre 2.12.0
• Labels: arm
• Environment: Arch: aarch64 (client)
• Severity: 3
• Assignee: Xinliang Liu
• Reporter: Jian Yu

Description

sanity test 45 failed on ARM clients as follows:

== sanity test 45: osc io page accounting ============================================================ 10:25:27 (1542191127)
vm.dirty_writeback_centisecs = 0
vm.dirty_writeback_centisecs = 0
vm.dirty_ratio = 50
vm.dirty_background_ratio = 25
executing "echo blah > /mnt/lustre/f45"
before 0, after 0
 sanity test_45: @@@@@@ FAIL: write wasn't cached 

Maloo report: https://testing.whamcloud.com/test_sets/d4705c44-e801-11e8-b67f-52540065bddc

Issue Links

• is related to: LU-15226 sanity test_45: FAIL: write wasn't cached
• is related to: LU-10300 Can the Lustre 2.10.x clients support 64K kernel page?

Comments

[James A Simmons]

Fix landed.

[Gerrit Updater]

"Oleg Drokin <[e-mail]>" merged in patch https://review.whamcloud.com/c/fs/lustre-release/+/55403/
Subject: LU-11671 tests: re-enable sanity test_45 for aarch64
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 16ac1c2641e8020bb9428ab671c32333c2efe3a1

[Xinliang Liu]

Hi James, thanks for the info. Let's see the test result of https://review.whamcloud.com/c/fs/lustre-release/+/55403.

 

[Gerrit Updater]

"xinliang <[e-mail]>" uploaded a new patch: https://review.whamcloud.com/c/fs/lustre-release/+/55403
Subject: LU-11671 tests: enable sanity test 45 for arm
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 522a4ae05e40342759b1a95517f3a303386f55bb

[James A Simmons]

A patch for https://jira.whamcloud.com/browse/LU-17733 has landed.  Xinliang can you see if this fixes this failure on ARM.

[James A Simmons]

This bug can also appear on x86 (LU-15226)

[Xinliang Liu]

Reproduced in recent test: https://testing.whamcloud.com/test_sets/531aa74e-3d10-47f3-9336-992f48db4d9b

Error: 'write wasn't cached'
Failure Rate: 0.00% of most recent 95 runs, 5 skipped (all branches)OST 1, OST 2, OST 3, OST 4, OST 5, OST 6, OST 7 (trevis-210vm1)sanity.test_45.debug_log.trevis-210vm1.1631877802.log [ Download | Show | Preview 50 ]
sanity.test_45.dmesg.trevis-210vm1.1631877802.log [ Download | Show | Preview 50 ]
MDS 1, MDS 3 (trevis-210vm2)sanity.test_45.debug_log.trevis-210vm2.1631877802.log [ Download | Show | Preview 50 ]
sanity.test_45.dmesg.trevis-210vm2.1631877802.log [ Download | Show | Preview 50 ]
MDS 2, MDS 4 (trevis-210vm3)sanity.test_45.debug_log.trevis-210vm3.1631877802.log [ Download | Show | Preview 50 ]
sanity.test_45.dmesg.trevis-210vm3.1631877802.log [ Download | Show | Preview 50 ]
Client 1 (trevis-79vm13)sanity.test_45.debug_log.trevis-79vm13.1631877802.log [ Download | Show | Preview 50 ]
sanity.test_45.dmesg.trevis-79vm13.1631877802.log [ Download | Show | Preview 50 ]
sanity.test_45.test_log.trevis-79vm13.log [ Download | Show | Preview 250 ]
== sanity test 45: osc io page accounting ====================================================================================== 11:23:16 (1631877796)
vm.dirty_writeback_centisecs = 0
vm.dirty_writeback_centisecs = 0
vm.dirty_ratio = 50
vm.dirty_background_ratio = 25
executing "echo blah > /mnt/lustre/f45"
before 0, after 65536
executing "> /mnt/lustre/f45"
before 65536, after 0
executing "echo blah > /mnt/lustre/f45"
before 0, after 0
 sanity test_45: @@@@@@ FAIL: write wasn't cached 
  Trace dump:
  = /usr/lib64/lustre/tests/test-framework.sh:6299:error()
  = /usr/lib64/lustre/tests/sanity.sh:5399:test_45()
  = /usr/lib64/lustre/tests/test-framework.sh:6603:run_one()
  = /usr/lib64/lustre/tests/test-framework.sh:6650:run_one_logged()
  = /usr/lib64/lustre/tests/test-framework.sh:6491:run_test()
  = /usr/lib64/lustre/tests/sanity.sh:5409:main()
Dumping lctl log to /autotest/autotest-2/2021-09-17/lustre-reviews_custom_83181_1_102_3240f885-4d31-402f-a8f0-9e13a3c88f88/sanity.test_45.*.1631877802.log
CMD: trevis-210vm1,trevis-210vm2,trevis-210vm3,trevis-79vm13.trevis.whamcloud.com,trevis-79vm14 /usr/sbin/lctl dk > /autotest/autotest-2/2021-09-17/lustre-reviews_custom_83181_1_102_3240f885-4d31-402f-a8f0-9e13a3c88f88/sanity.test_45.debug_log.\$(hostname -s).1631877802.log;
		dmesg > /autotest/autotest-2/2021-09-17/lustre-reviews_custom_83181_1_102_3240f885-4d31-402f-a8f0-9e13a3c88f88/sanity.test_45.dmesg.\$(hostname -s).1631877802.log
vm.dirty_writeback_centisecs = 500
vm.dirty_background_ratio = 10
vm.dirty_ratio = 30

[Jian Yu]

The test was added into ALWAYS_EXCEPT list in patch https://review.whamcloud.com/33652 .