[OpenAFS] Max number of files in volume?
Horst Birthelmer
horst@riback.net
Mon, 17 May 2004 12:17:55 +0200
On Monday, May 17, 2004, at 12:04 PM, Hamie wrote:
> Horst Birthelmer wrote:
>
>>
>> On Thursday, May 13, 2004, at 01:48 PM, Hamie wrote:
>>
>>> Tino Schwarze wrote:
>>>
>>>> On Thu, May 13, 2004 at 11:59:04AM +0100, Hamie wrote:
>>>>
>>>>
>>>>>> Just a thought ... Is your partition full??
>>>>>>
>>>>> No... 92%, but it's a 16GB filesystem (JFS on AIX 5.1).
>>>>>
>>>>>
>>>>> I did find this little bit of trouble though...
>>>>>
>>>>> r01ai01:/usr/afs/logs# bos salvage intl-mgw01.baplc.com -partition
>>>>> /vicepa -localauth
>>>>> bos: shutting down fs.
>>>>> Starting salvage.
>>>>> bos: waiting for salvage to complete.
>>>>> bos: salvage completed
>>>>> bos: restarting fs.
>>>>> br01ai01:/usr/afs/logs# ls
>>>>> br01ai01:/usr/afs/logs# cat SalvageLog
>>>>> @(#)Base configuration afs3.6 2.45
>>>>> 05/13/2004 11:54:40 STARTING AFS SALVAGER 2.4
>>>>> (/usr/afs/bin/salvager -f /vicepa)
>>>>> 05/13/2004 11:54:40 Unable to read block 1, partition
>>>>> 05/13/2004 11:54:40 Unable to read superblock on /proc.
>>>>> 05/13/2004 11:54:40 SALVAGING FILE SYSTEM PARTITION /vicepa
>>>>> (device=vicepa)
>>>>> 05/13/2004 11:54:40 ***Forced salvage of all volumes on this
>>>>> partition***
>>>>> 05/13/2004 11:54:49 File system /dev/rvicepa is in a bad state.
>>>>> 05/13/2004 11:54:49 Call your IBM representative.
>>>>> Unable to get inodes for "vicepa"; not salvaged
>>>>>
>>>>
>>>> I don't know anything about JFS, but is it possible that the
>>>> partition
>>>> is out of inodes?
>>>>
>>>>
>>>
>>> unfortunately (?) not..
>>>
>>> /dev/vicepa 15597568 1302472 92% 270 1% /vicepa
>>> /dev/vicepb 15597568 15066904 4% 587 1% /vicepb
>>>
>>> Only 1% inodes used...
>>>
>> Is this a namei fileserver??
>> If yes, try to enter /vicepa and do something in the directory.
>> Copy some files or ... whatever just for making shure your filesystem
>> is accessible.
>>
>
> The answer was sort of... AFS could read & write happily to existing
> volumes... But the filesystem appeared to be hosed somewhere between
> the JFS itself & the disk...
>
> Probably due to a power outage last week on the ESS that holds the
> actual disk. I had to fsck the filesystems (/vicepa & /vicepb) before
> they'd mount, and I'm running a salvage now, but it looks like it's
> come back. Very strange, because it was only the JFS filesystems on
> that volumegroup that were affected. The jfs2 filesystems were fine...
> There are a whole lot (9) of empty volume files in the /vicepa
> filesystem though... They look like they're from attempts to move
> volumes late last week...
>
What do you actually mean by "empty volume files" ??
In a namei fileserver the *.vol files in the root directory of the
partition (vicep*) don't contain any 'real' data from the volume.
Horst