[LU-17781] sanity-pcc test_33 Inconsistent SOM data on rhel9.3

• Type: Bug
• Resolution: Unresolved
• Priority: Minor
• Severity: 3
• Assignee: WC Triage
• Reporter: Qian Yingjin

Description

The sanity-pcc/test_33 on master rhel9.3 kernel failed due to the inconsistent LSOM data:
https://testing.whamcloud.com/test_sessions/5140697d-8ada-4d8f-b39c-3f63085c5cc5

After analyzed, found that it is a common bug in the Lustre. It can be reproduced by the following test scripts:

disable_opencache
touch $file2 || error "failed to touch $file"
$TRUNCATE $file2 $((1048576 * 4)) || error "Truncate $file2 failed"
$MULTIOP $file oc || error "failed to readonly open $file"
$TRUNCATE $file2 $((1048576 * 5)) || error "Truncate $file failed"
$MULTIOP $file oc || error "failed to readonly open $file"
cat $file > /dev/null
$TRUNCATE $file2 $((4 * 1048576)) || error "Truncate $file failed"
$LFS getsom $file2
$LFS getsom $file
#$LCTL set_param ldlm.namespaces.*-ffff8955ed35d800.lru_size=clear
$MULTIOP $file oc || error "failed to readonly open $file"
$LFS getsom $file
#$LCTL set_param ldlm.namespaces.*-ffff8955ed35d800.lru_size=clear
$LFS getsom $file

 

The output:

llite.lustre-ffff8955c4a6e000.opencache_threshold_count=off
llite.lustre-ffff8955ed35d800.opencache_threshold_count=off
file: /mnt/lustre2/tfile.som size: 4194304 blocks: 1 flags: 4
file: /mnt/lustre/tfile.som size: 4194304 blocks: 1 flags: 4
file: /mnt/lustre/tfile.som size: 5242880 blocks: 1 flags: 4
file: /mnt/lustre/tfile.som size: 5242880 blocks: 1 flags: 4

 

The SOM size should be 4MiB, but the final SOM size was 5MiB...

Issue Links

• is related to: LU-17978 sanity-pcc test_33: FAIL: /mnt/lustre/myfile.doc expected pcc state: none, but got: readonly