Hi,
I am having a quota issue with Lustre. Here are the default limits
set for an account -
/home 0 400000 450000 189 100000 100000
This sets a quota of about 300 MB or so. After reading the
documenatation, i throught this would have set ~4 G or so, but it is one
zero less. The quota limit is working though.
But today i found the following issue with two accounts. The
account goes over quota showing higher block utilization, while the real
space occupied by the account is only a few megabytes on the filesystem.
Here is an example -
Filesystem blocks quota limit grace files quota limit
grace
/home 307796 600000 650000 221 100000 100000
home-MDT0000_UUID
316 0 102400 221 0 5000
home-OST0000_UUID
307480 0 409600> du -sh ../username/
3.8M ../username/
In the Lustre server, there are quite a few of the following error
messages -
Jul 24 01:31:09 lustre-3ware kernel: LustreError:
3659:0:(fsfilt-ldiskfs.c:1962:fsfilt_ldiskfs_dquot()) Skipped 11 previous
similar messages
Jul 24 01:31:09 lustre-3ware kernel: LustreError:
3659:0:(quota_master.c:194:lustre_dqget()) can''t read dquot from admin
quotafile! (rc:-5)
Jul 24 01:31:09 lustre-3ware kernel: LustreError:
3659:0:(quota_master.c:194:lustre_dqget()) Skipped 11 previous similar
messages
Jul 24 01:31:09 lustre-3ware kernel: LustreError:
3659:0:(ldlm_lib.c:1554:target_handle_dqacq_callback()) dqacq failed!
(rc:-5)
Jul 24 01:31:09 lustre-3ware kernel: LustreError:
3537:0:(quota_context.c:422:dqacq_completion()) acquire qunit got error!
(rc:-5)
Regards
Balagopal
On Tue, 24 Jul 2007, Balagopal Pillai wrote:
Hi,
There is one more line in the error messages i missed to include
last time -
Jul 24 01:31:19 lustre-3ware kernel: LustreError:
3657:0:(fsfilt-ldiskfs.c:1962:fsfilt_ldiskfs_dquot()) operate dquot
before it''s enabled!
Thanks
Balagopal
> Hi, >
> I am having a quota issue with Lustre. Here are the default limits
> set for an account -
> /home 0 400000 450000 189 100000 100000
>
> This sets a quota of about 300 MB or so. After reading the
> documenatation, i throught this would have set ~4 G or so, but it is one
> zero less. The quota limit is working though.
>
> But today i found the following issue with two accounts. The
> account goes over quota showing higher block utilization, while the real
> space occupied by the account is only a few megabytes on the filesystem.
> Here is an example -
>
> Filesystem blocks quota limit grace files quota limit
> grace
> /home 307796 600000 650000 221 100000 100000
> home-MDT0000_UUID
> 316 0 102400 221 0 5000
> home-OST0000_UUID
> 307480 0 409600
> > du -sh ../username/
> 3.8M ../username/
>
>
> In the Lustre server, there are quite a few of the following error
> messages -
>
> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
> 3659:0:(fsfilt-ldiskfs.c:1962:fsfilt_ldiskfs_dquot()) Skipped 11 previous
> similar messages
> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
> 3659:0:(quota_master.c:194:lustre_dqget()) can''t read dquot from
admin
> quotafile! (rc:-5)
> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
> 3659:0:(quota_master.c:194:lustre_dqget()) Skipped 11 previous similar
> messages
> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
> 3659:0:(ldlm_lib.c:1554:target_handle_dqacq_callback()) dqacq failed!
> (rc:-5)
> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
> 3537:0:(quota_context.c:422:dqacq_completion()) acquire qunit got error!
> (rc:-5)
>
>
>
> Regards
> Balagopal
>
> _______________________________________________
> Lustre-discuss mailing list
> Lustre-discuss@clusterfs.com
> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss
>
Hi,
There are a few further messages on the Lustre server
like this -
Jul 24 08:25:35 lustre-3ware kernel: Lustre: Found inode with zero
generation or link -- this may indicate disk corruption (inode:
10251612/839130903, link 0, count 1)
Jul 24 08:25:35 lustre-3ware last message repeated 2 times
Jul 24 08:25:35 lustre-3ware kernel: Lustre: Skipped 6 previous similar
messages
Jul 24 08:25:35 lustre-3ware kernel: Lustre: Found inode with zero
generation or link -- this may indicate disk corruption (inode:
10251612/839130903, link 0, count 1)
Jul 24 08:25:35 lustre-3ware kernel: Lustre: Skipped 6 previous similar
messages
Jul 24 09:09:10 lustre-3ware kernel: Lustre: Found inode with zero
generation or link -- this may indicate disk corruption (inode:
10251612/839130903, link 0, count 1)
Jul 24 09:09:10 lustre-3ware kernel: Lustre: Found inode with zero
generation or link -- this may indicate disk corruption (inode:
10251612/839130903, link 0, count 1)
Jul 24 09:09:10 lustre-3ware kernel: Lustre: Skipped 3 previous similar
messages
Jul 24 09:09:10 lustre-3ware kernel: Lustre: Found inode with zero
generation or link -- this may indicate disk corruption (inode:
10251612/839130903, link 0, count 1)
Jul 24 09:09:10 lustre-3ware kernel: Lustre: Skipped 3 previous similar
messages
Jul 24 09:09:10 lustre-3ware kernel: Lustre: Found inode with zero
generation or link -- this may indicate disk corruption (inode:
10251612/839130903, link 0, count 1)
This is preceded by lots of the following error messages -
Jul 24 01:31:24 lustre-3ware kernel: LustreError:
3654:0:(ldlm_lib.c:1554:target_handle_dqacq_callback()) dqacq failed!
(rc:-5)
The data on the volumes is fine. Is the disk corruption error
related to the quota problem or is this a new issue? The 3ware raid
volume is working fine. Thanks very much.
Regards
Balagopal
Balagopal Pillai wrote:> On Tue, 24 Jul 2007, Balagopal Pillai wrote:
> Hi,
>
> There is one more line in the error messages i missed to include
> last time -
>
> Jul 24 01:31:19 lustre-3ware kernel: LustreError:
> 3657:0:(fsfilt-ldiskfs.c:1962:fsfilt_ldiskfs_dquot()) operate dquot
> before it''s enabled!
>
> Thanks
> Balagopal
>
>
>> Hi, >
>> I am having a quota issue with Lustre. Here are the default
limits
>> set for an account -
>> /home 0 400000 450000 189 100000 100000
>>
>> This sets a quota of about 300 MB or so. After reading the
>> documenatation, i throught this would have set ~4 G or so, but it is
one
>> zero less. The quota limit is working though.
>>
>> But today i found the following issue with two accounts. The
>> account goes over quota showing higher block utilization, while the
real
>> space occupied by the account is only a few megabytes on the
filesystem.
>> Here is an example -
>>
>> Filesystem blocks quota limit grace files quota limit
>> grace
>> /home 307796 600000 650000 221 100000 100000
>> home-MDT0000_UUID
>> 316 0 102400 221 0 5000
>> home-OST0000_UUID
>> 307480 0 409600
>>
>>> du -sh ../username/
>>>
>> 3.8M ../username/
>>
>>
>> In the Lustre server, there are quite a few of the following error
>> messages -
>>
>> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
>> 3659:0:(fsfilt-ldiskfs.c:1962:fsfilt_ldiskfs_dquot()) Skipped 11
previous
>> similar messages
>> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
>> 3659:0:(quota_master.c:194:lustre_dqget()) can''t read dquot
from admin
>> quotafile! (rc:-5)
>> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
>> 3659:0:(quota_master.c:194:lustre_dqget()) Skipped 11 previous similar
>> messages
>> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
>> 3659:0:(ldlm_lib.c:1554:target_handle_dqacq_callback()) dqacq failed!
>> (rc:-5)
>> Jul 24 01:31:09 lustre-3ware kernel: LustreError:
>> 3537:0:(quota_context.c:422:dqacq_completion()) acquire qunit got
error!
>> (rc:-5)
>>
>>
>>
>> Regards
>> Balagopal
>>
>> _______________________________________________
>> Lustre-discuss mailing list
>> Lustre-discuss@clusterfs.com
>> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss
>>
>>
>
> _______________________________________________
> Lustre-discuss mailing list
> Lustre-discuss@clusterfs.com
> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss
>
On Jul 24, 2007 09:45 -0300, Balagopal Pillai wrote:> There are a few further messages on the Lustre server > like this - > > Jul 24 09:09:10 lustre-3ware kernel: Lustre: Found inode with zero > generation or link -- this may indicate disk corruption (inode: > 10251612/839130903, link 0, count 1) > > The data on the volumes is fine. Is the disk corruption error > related to the quota problem or is this a new issue? The 3ware raid > volume is working fine. Thanks very much.This is a spurious error message that has been removed in the upcoming releases. Cheers, Andreas -- Andreas Dilger Principal Software Engineer Cluster File Systems, Inc.
Details in CFS bug 10419 -----Original Message----- From: Andreas Dilger [mailto:adilger@clusterfs.com] Sent: 24 July 2007 16:58 To: Balagopal Pillai Cc: lustre-discuss@clusterfs.com Subject: Re: [Lustre-discuss] lustre quota problem On Jul 24, 2007 09:45 -0300, Balagopal Pillai wrote:> There are a few further messages on the Lustre server > like this - > > Jul 24 09:09:10 lustre-3ware kernel: Lustre: Found inode with zero > generation or link -- this may indicate disk corruption (inode: > 10251612/839130903, link 0, count 1) > > The data on the volumes is fine. Is the disk corruptionerror> related to the quota problem or is this a new issue? The 3ware raid > volume is working fine. Thanks very much.This is a spurious error message that has been removed in the upcoming releases. Cheers, Andreas -- Andreas Dilger Principal Software Engineer Cluster File Systems, Inc.