[OpenAFS] trouble deleting corrupt volume

[Wes Chow]

Anne.Salemme@Dartmouth.edu wrote:
> what does the vldb think about it? can you show us the error messages
> when you try to clean it up? someone will have some good ideas...
>
> anne


The volume isn't listed as being on that server in the vldb:

wchow@ein:~/vols$ vos examine dfs.200510
dfs.200510                        536942164 RW  107376809 K  On-line
    fs1.dev.in.athenacr.com /vicepb
    RWrite  536942164 ROnly  536943971 Backup          0
    MaxQuota  200000000 K
    Creation    Wed Sep 28 11:57:12 2005
    Copy        Wed Dec 14 14:30:16 2005
    Backup      Never
    Last Update Tue Dec  6 17:27:20 2005
    0 accesses in the past day (i.e., vnode references)

    RWrite: 536942164     ROnly: 536942165
    number of sites -> 2
       server fs1.dev.in.athenacr.com partition /vicepb RW Site
       server hippo.dev.in.athenacr.com partition /vicepb RO Site


And when I try to zap it:

wchow@ein:~/vols$ vos zap fs0.dev /vicepa dfs.200510
Warning: Entry for volume number 536942164 exists in VLDB (but we're
zapping it anyway!)Could not start transaction on volume 536942164
Volume needs to be salvaged
Error in vos zap command.
Volume needs to be salvaged


I'm not entirely sure how I did this.  I *think* it's because I took
what may have already been an iffy dfs.200510.readonly volume on
fs0.dev/vicepa and converted it to a RW volume, while the RW volume on
fs1.dev still existed.



> Quoting Wes Chow <wchow@athenacr.com>:
> 
>>
>> I'm trying to get rid of a corrupt volume on a server, but attempts to
>> zap it have seemed to leave it in an inconsistent state:
>>
>> wchow@ein:~/vols$ vos listvol fs0.dev
>> Total number of volumes on server fs0.dev partition /vicepa: 8
>> dfs.200504.readonly               536941570 RO   86680945 K On-line
>> dfs.200505.readonly               536941573 RO   90877122 K On-line
>> dfs.200506.readonly               536941576 RO  121191000 K On-line
>> dfs.200507.readonly               536941579 RO   97423014 K On-line
>> dfs.200508.readonly               536941582 RO  106688000 K On-line
>> dfs.200509.readonly               536941585 RO  109427632 K On-line
>> dfs.200511.readonly               536941588 RO  100987584 K On-line
>> **** Could not attach volume 536942164 ****
>>
>>
>> When I salvage the volume, I get this in the log:
>>
>> fs0:/vicepa/AFSIDat# tail -f /var/log/openafs/SalvageLog
>> 06/05/2006 11:50:09 Found 0 link count file
>> /vicepa/AFSIDat/I=/INF+U/+/+/82+++M6.
>> 06/05/2006 11:50:09 Found 0 link count file
>> /vicepa/AFSIDat/I=/INF+U/+/+/A2+++Q6.
>> 06/05/2006 11:50:09 Found 0 link count file
>> /vicepa/AFSIDat/I=/INF+U/+/+/C2+++U6.
>> 06/05/2006 11:50:09 Found 0 link count file
>> /vicepa/AFSIDat/I=/INF+U/+/+/=+++22.
>> 06/05/2006 11:50:09 1 nVolumesInInodeFile 28
>> 06/05/2006 11:50:09 SALVAGING VOLUME 536942164.
>> 06/05/2006 11:50:09 dfs.200510 (536942164) updated 12/06/2005 17:27
>> 06/05/2006 11:50:09 totalInodes 116
>> 06/05/2006 11:50:09 Found 112 orphaned files and directories (approx.
>> 37020063 KB)
>> 06/05/2006 11:50:09 Salvaged dfs.200510 (536942164): 112 files, 37020063
>> blocks
>>
>>
>> However, it remains unattachable (and unzappable).
>>
>> I would like to remove the volume from the server so that I can recreate
>> it from our backup.  I can't seem to do this through the regular
>> utilities, however... any suggestions?
>>
>>
>> Thanks,
>> Wes

-- 

Athena Capital Research LLC
http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xA5CA6644
Key fingerprint = FDE5 21D8 9D8B 386F 128F  DF52 3F52 D582 A5CA 6644