[Ocfs2-users] OCFS2 showing "No space left on device" on a device with free space

[Herman L]

>Em Tue, 6 Dec 2011 17:58:15 +0800
>"Ramappa, Ravi (NSN - IN/Bangalore)" <ravi.ramappa at nsn.com> escreveu:
>
>> Hi ,
>>
>> I am getting the error "No space left on device" on a device with free
>> space which is ocfs2 filesystem.
>> Additional information is as below,
>>
>> [root at sai93 staging]# debugfs.ocfs2 -n -R "stats" /dev/sdb1 | grep -i
>> "Cluster Size"
>>         Block Size Bits: 12   Cluster Size Bits: 15
>>
>> [root at sai93 ~]# cat /etc/redhat-release
>> Red Hat Enterprise Linux Server release 5.3 (Tikanga)
>>
>> [root at sai93 ~]# uname -r
>> 2.6.18-128.el5
>>
>> [root at sai93 ~]# rpm -qa| grep ocfs
>> ocfs2console-1.4.4-1.el5
>> ocfs2-2.6.18-128.el5-1.4.7-1.el5
>> ocfs2-tools-1.4.4-1.el5
>>
>> [root at sai93 ~]# cat /etc/ocfs2/cluster.conf
>> node:
>>         ip_port = 7777
>>         ip_address = 10.71.10.93
>>         number = 0
>>         name = sai93
>>         cluster = ocfs2
>> node:
>>         ip_port = 7777
>>         ip_address = 10.71.10.94
>>         number = 1
>>         name = sai94
>>         cluster = ocfs2
>> cluster:
>>         node_count =2
>>         name = ocfs2
>>
>> [root at sai93 ~]#
>>
>> The only way I overcome this error is by unmounting the ocfs2
>> filesystem, doing a fsck followed by server reboot.
>>
>> 988  fsck.ocfs2 /dev/sdb1
>>   989  mount -a
>>   990  df -h
>>   991  touch /mnt/staging/test
>>   992  man fsck.ocfs2
>>   993  umount /mnt
>>   994  service ocfs2 stop
>>   995  dmesg
>>   996  reboot
>>
>> After restarting the servers  I have executed the stat_sysdir.sh
>> script, the output of which is attached herwith. Kindly confirm
>> whether I need to do any changes to ocfs2 configuration , to avoid
>> the above error.
>>
>>  <<output.zip>>
>>
>> Thanks & Regards,
>>
>> Ravi  Ramappa
>
>Hello,
>
>This is a known issue and was fixed in ocfs2 1.6 series by enabling the
>discontig-bg feature.
>
>If you are not using DRDB you can upgrade your kernel and ocfs2-tools
>using the ULEK 5.5 series, provided by oracle at
>http://public-yum.oracle.com/ .
>
>Regards,
>Sérgio

We are also having a similar issue, but we are using DRBD.  We are set at the minimum 2 node slots, so I can't reduce those any more.

Is there any other good way to fix this problem, at least for a while?  What about ways to preemptively clear some fragmentation before we get the dreaded error?

I think moving files off the filesystem and back works somewhat, but how can I tell which files to move would have the most effect?

Thanks,
Herman
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://oss.oracle.com/pipermail/ocfs2-users/attachments/20111207/b9c4846e/attachment.html