--On Tuesday, January 04, 2005 12:53:58 PM -0500 Derrick J Brashear
<[EMAIL PROTECTED]> wrote:

>> So far, the results using memcache are: good (but I haven't done any
>> further timing tests).  The results using disk cache are: bad (some
>> silent truncation/corruption using Derrick's suggested patch to rx.c
>> from 2004-12-22
> 
> are you using the same rx patch with memcache also?

Yes.

I haven't tried these variations yet:

- keep rx patch, but with larger memcache

- no rx patch, probably with larger memcache

Would trying those help?

Past results:

- no rx patch, disk cache = oops (usually)

- rx patch, disk cache = silent corruption (occasionally)

- rx patch, with default memcache = slow, but no oops or corruption

This is all with 1.3.77 on FC3 with FC3 2.6.9-1.681 kernel (only
change: export syscall table).

Thanks,
Rudy
_______________________________________________
OpenAFS-info mailing list
[email protected]
https://lists.openafs.org/mailman/listinfo/openafs-info

Reply via email to