arla failure

[Reinhold Bader]

Thomas Jordan wrote:

> Dear Reinhold,
>
> I have noticed this occur when my afs tokens have expired. My 
> keyserver sets a token lifetime and unless I renew the token within 
> that period the token dies. When I reautheniticate, my permissions are 
> fine. 

That unfortunately does not seem to be the problem. The following 
sequence of events is typical

* start arla
* klog
* copy a large file (ca. 120 MB) from AFS (cache is set to 200 MB min, 
500 MB max)
* after copying, stat to AFS is impossible
* tokens shows that token is still valid
* unlog
* tokens
* stat to AFS is still impossible

However I have found an interesting new kernel message (local 
filesystems are reiserfs)

Nov  6 10:00:04 lrzpc12103 kernel: PAP-5710: reiserfs_paste_into_item: 
entry or
pasted byte ([153304 153316 0xfa001 IND]) exists<4>PAP-5710: 
reiserfs_paste_into_item: entry or pasted byte ([153304 153316 0x50e6001 
IND]) exists<4>vs-825: reiserfs_get_block: [153304 153316 0x94c8001 
UNKNOWN] should not be found

Do arla and reiserfs use the VFS in an inconsistent manner??

>
>
> The next time that happens run /usr/arla/bin/tokens to see what tokens 
> Arla holds:
>
>     [owen-meany:/usr/arla/bin] jordant% unlog
>     [owen-meany:/usr/arla/bin] jordant% ./tokens
>
>     Tokens held by Arla:
>
>        --End of list--
>     [owen-meany:/usr/arla/bin] jordant% kalog
>     Getting ticket for jordant at fnal.gov
>     Password:
>     kerberos-iv/udp unknown service, using default port 750
>     [owen-meany:/usr/arla/bin] jordant% ./tokens
>
>     Tokens held by Arla:
>     
>     User's (AFS ID 8998) tokens for afs at fnal.gov [Expires Nov  5 21:33]
>        --End of list--
>     [owen-meany:/usr/arla/bin] jordant% date
>     Mon Nov  5 10:53:49 CST 2001
>     [owen-meany:/usr/arla/bin] jordant%
>
> Note that the afs token that I just received expires in <12 hours.
>
> Best regards,
>     Tom
>
>
> On Monday, November 5, 2001, at 04:06 AM, Reinhold Bader wrote:
>
>> Hello all,
>>
>> arla 0.35.6 fails some time after initialization on
>> kernels 2.4.10, 2.4.13. One gets kernel log messages
>>
>> Nov  5 10:52:39 lrzpc12103 kernel: xfs_readdir: data not valid: -4
>> Nov  5 10:52:49 lrzpc12103 kernel: xfs_readdir: data not valid: -2
>>
>>
>> and it is not possible to copy any data from AFS any more. sometimes
>> it is not even possible to stat AFS directories.
>>
>> Any hints?
>>
>>
>>
>>                                               Best regards
>>
>> -- Dr. Reinhold Bader
>>
>> Leibniz-Rechenzentrum, Abt. Benutzerbetreuung    | Tel. +4989 289 28825
>> Barerstr. 21, 80333 Muenchen            | email Bader at lrz.de
>>
>>
>>
>>
> Thomas Jordan
> Fermi National Accelerator Laboratory
> PO Box 500, MS 226 WH15W
> Batavia, Il
> 60510-0500
> ofc:630.840.4035 fax:630.840.8248
>


-- 
 Dr. Reinhold Bader

 Leibniz-Rechenzentrum, Abt. Benutzerbetreuung	| Tel. +4989 289 28825
 Barerstr. 21, 80333 Muenchen			| email Bader at lrz.de