[Lustre-discuss] Question about quotas.
Niu, Yawei
yawei.niu at intel.com
Tue Jan 22 01:26:43 PST 2013
Looks your OST 2 is administratively disabled, you'd activate all your OSTs if you want to get accurate quota usage. Please refer to manual to see how activate/deactivate OSTs.
Thanks
- Niu
From: ???????? ????????? ???????? <theodoros.stylianos.kondylis at gmail.com<mailto:theodoros.stylianos.kondylis at gmail.com>>
Date: Friday, January 18, 2013 5:40 PM
To: Yawei Niu <yawei.niu at intel.com<mailto:yawei.niu at intel.com>>
Subject: Re: [Lustre-discuss] Question about quotas.
Thank you all for your information.
I tried your proposals, and I found the following ::
clie1# lfs check servers
jtest1-MDT0000-mdc-ffff88033d143400 active.
jtest1-OST0001-osc-ffff88033d143400 active.
jtest1-OST0003-osc-ffff88033d143400 active.
jtest1-OST0000-osc-ffff88033d143400 active.
jtest1-OST0002-osc-ffff88033d143400 active.
But when I do a ::
# lfs quota /lustre/jtest1/
Disk quotas for user root (uid 0):
Filesystem kbytes quota limit grace files quota limit grace
/lustre/jtest1/ [15751753680] 30786325577728 31555983717171 - [1280] 10000 11000 -
Some errors happened when getting quota info. Some devices may be not working or deactivated. The data in "[]" is inaccurate.
Disk quotas for group root (gid 0):
Filesystem kbytes quota limit grace files quota limit grace
/lustre/jtest1/ [15751753680] 30786325577728 31555983717171 - [1280] 10000 11000 -
Some errors happened when getting quota info. Some devices may be not working or deactivated. The data in "[]" is inaccurate.
At the same time in the mds I see ::
mds2# dmesg | tail
LustreError: 7116:0:(quota_ctl.c:463:lov_quota_ctl()) ost 2 is inactive
LustreError: 7116:0:(quota_ctl.c:463:lov_quota_ctl()) Skipped 1 previous similar message
But on the OSS1 (responsible for ost2) the OST2 md array is mounted.
How I can check further about the ost's status ? Since the OST's md_array is mounted, shouldn't that be active ?
I even try through heartbeat a failover from OSS1 to OSS2 but the ost2 remained inactive ( despite the ost2's md_array being mounted ).
If you have any idea or indication to lustre commands/files that would be very helpful.
On Thu, Jan 17, 2013 at 10:01 AM, Niu, Yawei <yawei.niu at intel.com<mailto:yawei.niu at intel.com>> wrote:
Perhaps quota wasn't able to turned on for some OSTs, did you see any error message (in syslog) when starting MDT & OSTs.
From: ???????? ????????? ???????? <theodoros.stylianos.kondylis at gmail.com<mailto:theodoros.stylianos.kondylis at gmail.com>>
Date: Wednesday, January 16, 2013 2:26 AM
To: "lustre-discuss at lists.lustre.org<mailto:lustre-discuss at lists.lustre.org>" <lustre-discuss at lists.lustre.org<mailto:lustre-discuss at lists.lustre.org>>
Subject: [Lustre-discuss] Question about quotas.
Hello to everyone.
We have a lustre (v1.8.4) test cluster with 2 MDSs, 2OSSs and 4 Clients.
I am experimenting with the quotas but something seems to not work.
In first I did a
cli1# lfs quotacheck /lustre
cli1# lfs quotaon -ug /lustre
cli1# lfs quota /lustre
To get ::
Disk quotas for user root (uid 0):
Filesystem kbytes quota limit grace files quota limit grace
/lustre/jtest1/ [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.
Disk quotas for group root (gid 0):
Filesystem kbytes quota limit grace files quota limit grace
/lustre/jtest1/ [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.
So I tried ::
mds2# cat /proc/fs/lustre/lquota/jtest1-MDT0000/quota_type
>> ug3
oss1# cat /proc/fs/lustre/lquota/jtest1-OST000{1,3}/quota_type
>> ug3
>> ug3
oss2# cat /proc/fs/lustre/lquota/jtest1-OST000{0,2}/quota_type
>> ug3
>> ug3
Then I tried ::
cli1# lfs quotaon -ugf /lustre
cli1# lfs quota /lustre
>>user quotas are not enabled.
>>group quotas are not enabled.
mds2# cat /proc/fs/lustre/lquota/jtest1-MDT0000/quota_type
>> 3
oss1# cat /proc/fs/lustre/lquota/jtest1-OST000{1,3}/quota_type
>> 3
>> 3
oss2# cat /proc/fs/lustre/lquota/jtest1-OST000{0,2}/quota_type
>> 3
>> 3
And then again ::
cli1# lfs quotaon -ug /lustre
cli1# lfs quota /lustre
>> Disk quotas for user root (uid 0):
Filesystem kbytes quota limit grace files quota limit grace
>> /lustre/jtest1/ [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.
>> Disk quotas for group root (gid 0):
Filesystem kbytes quota limit grace files quota limit grace
>> /lustre/jtest1/ [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.
mds2# cat /proc/fs/lustre/lquota/jtest1-MDT0000/quota_type
>> ug3
oss1# cat /proc/fs/lustre/lquota/jtest1-OST000{1,3}/quota_type
>> ug3
>> ug3
oss2# cat /proc/fs/lustre/lquota/jtest1-OST000{0,2}/quota_type
>> ug3
>> ug3
So I would like to make a few questions in case anyone knows.
First of all is this a known issue, or am I doing something wrong here? I am forcing the quotaon and that reacts as if I did a quotaoff.
Furthermore I would like to ask about this number 3 in the quota_type files in case anyone knows what does it mean and why is this necessary.
Finally if I am not doing something wrong here, is there a way to fix this ?
Thank you in advance for your time and any replies/guidance/directions.
Stelios.
More information about the lustre-discuss
mailing list