[LU-150] inaccurate quota output after the recovery Created: 21/Mar/11  Updated: 28/Jun/11  Resolved: 28/Mar/11

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: Lustre 1.8.6
Fix Version/s: Lustre 1.8.6

Type: Bug Priority: Blocker
Reporter: Shuichi Ihara (Inactive) Assignee: Niu Yawei (Inactive)
Resolution: Won't Fix Votes: 0
Labels: None

Severity: 3
Rank (Obsolete): 10124

 Description   

We can't see correct quota output after the recovery. Once run quotacheck again, the 'lfs quota' shows right output. Here is quick reproducer. We confirmed same situation happens on both lustre-1.8.4 and lustre-1.8.5.

0. Normal case
It's no problem.

[test@r03 test]$ lfs quota -u test /mnt/lustre
Disk quotas for user test (uid 501):
Filesystem kbytes quota limit grace files quota limit grace
/mnt/lustre 4 0 0 - 1 0 0 -

1. 'lfs quota' on during the recovery
Umount/mount an OST and occurs the recovery. The during the recovery, when we run 'lfs quota' command, the some data are inaccurate.

Disk quotas for user test (uid 501):
Filesystem kbytes quota limit grace files quota limit grace
/mnt/lustre [0] 0 0 - [0] 0 0 -
Some errors happened when getting quota info. Some devices may be not working or deactivated. The data in "[]" is inaccurate.

2. 'lfs quota' after the recovery

Even after finish the recovery, inaccurate output still can't fixed.

Disk quotas for user test (uid 501):
Filesystem kbytes quota limit grace files quota limit grace
/mnt/lustre [0] 0 0 - [0] 0 0 -
Some errors happened when getting quota info. Some devices may be not working or deactivated. The data in "[]" is inaccurate.

3. run quotacheck again

the correct output shows up once we run quotacheck again after the recovery.

Disk quotas for user test (uid 501):
Filesystem kbytes quota limit grace files quota limit grace
/mnt/lustre 4 0 0 - 1 0 0 -



 Comments   
Comment by Peter Jones [ 22/Mar/11 ]

Niu

Could you please look into this issue?

Thanks

Peter

Comment by Niu Yawei (Inactive) [ 22/Mar/11 ]

Hi, Ihara

This is because the quota on the umount/mount OST was not turned on, so lfs reported the usage might be inaccurate. You just need to run 'lfs quotaon' again to turn on quota on all OSTs after that.

Thanks

  • Niu
Comment by Shuichi Ihara (Inactive) [ 23/Mar/11 ]

Niu,

'lfs quotaon' helps to avoid 'lfs quotacheck' again after the recovery, but still need to run 'lfs quotaon' after the every recovery finish. The writing of ost.quota_type=ug to all OSTs, mdt.quota_type=ug to MDT are only a solution to keep the enabling the quota?

we think about the failover situation of this.

Ihara

Comment by Niu Yawei (Inactive) [ 23/Mar/11 ]

Hi, Ihara

Right, quota should not be automatically enabled after filesystem mounted, unless there are configuration like ost.quota_type, mdt/mdd.quota_type (I think it's the best way to auto turn on quota for lustre), or some auto-scripts to do it (like local fs usually does).

Thanks

  • Niu
Comment by Shuichi Ihara (Inactive) [ 28/Mar/11 ]

OK, thanks! I also tested again and confirmed this. It was not bug. I was sorry this stupid question. I've passed quota related information to the customer. Please close this ticket.

Ihara

Comment by Niu Yawei (Inactive) [ 28/Mar/11 ]

Thank you, Ihara. Set it as 'Won't fix'.

Generated at Sat Feb 10 01:04:15 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.