[OpenAFS] diskless client and Segmentation fault

[Info]

    Info wrote:
>> Hi list,
>>
>> starting afsd I will get a Segmentation fault, but all daemons started
>> and seem to work well.
>>
>> tested kernel- and openafs-versions with same results are:
>> linux 2.6.14.1 and openafs 1.4.0
>> suse-kernel 2.6.11.4-21.9 and openafs 1.3.78
>>
>> No problems with:
>> suse-kernel 2.4.21 and openafs 1.2.10
>>
>> The segmentation fault happens not with rw-exported and -mounted
>> root-directory.
>>
>> I start openafs with:
>> # modprobe libafs
>> # afsd -stat 300 -dcache 100 -daemons 2 -volumes 50 -logfile
>> /var/log/openafs/AFSlog -debug -memcache -verbose -dynroot
>> Segmentation fault
>>
>> the strace-log shows:
>> ...
>> afsd: Forking 2 background daemons.
>> afsd: Calling AFSOP_VOLUMEINFO: volume info file is '/VolumeItems'
>> afsd: Calling AFSOP_AFSLOG: volume info file is
>> '/var/log/openafs/AFSlog'
>> afsd: Calling AFSOP_CACHEINODE for each of the 100 files in ''
>> afsd: Calling AFSOP_GO with cacheSetTime = 0
>> SScall(137, 28, 100)=0 afsd: All AFS daemons started.
>> afsd: Forking trunc-cache daemon.
>> SScall(137, 28, 3)=0 +++ killed by SIGSEGV +++
>> afsd: My home cell is 'rz'
>> ParseCacheInfoFile: Opening cache info file '/etc/openafs/cacheinfo'...
>> ParseCacheInfoFile: Cache info file successfully parsed:
>>         cacheMountDir: '/afs'
>>         cacheBaseDir: '/var/cache/openafs'
>>         cacheBlocks: 64485
>> afsd: 100 inode_for_V entries at 0x80731c0, 400 bytes
>> SScall(137, 28, 17)=0 afsd: Forking rx listener daemon.
>> afsd: Forking rx callback listener.
>> SScall(137, 28, 0)=0
>>
>> /var/log/messages shows:
>> Nov 12 20:54:48 linux kernel: libafs: no version magic, tainting kernel.
>> Nov 12 20:54:48 linux kernel: Found system call table at 0xc0372780
>> (exported)
>> Nov 12 20:54:48 linux kernel: Starting AFS cache scan...Memory cache:
>> Allocating 100 dcache entries...found 0 non-empty cache files (0%).
>>
>> Please help, what can I do?
>> Ralf
>> _______________________________________________
>> OpenAFS-info mailing list
>> OpenAFS-info@openafs.org
>> https://lists.openafs.org/mailman/listinfo/openafs-info
Mike Polek schrieb:
> Hello,
>   I noticed that you use -dcache 100, rather than specifying
> blocks and/or chunksize. According to the documentation, this
> isn't recommended. Were you really intending an 800K cache size?
>
>   The parameters we use here are:
>
> afsd -memcache -blocks 65536 -chunksize 16 -stat 2800 -daemons 3
> -volumes 128 -nosettime
>
> That's the largest cache I've been able to get working correctly.
> You may want something smaller... but it seems to me that you may
> have gone for something a bit too small. Assuming your machine
> has enough memory, you can try the params above, or maybe
>
> -blocks 8192 -chunksize 12
>
> if you're strapped for memory...
>
> Just some suggestions. I know I had this kind of problem years ago.
> I believe I just solved it by tweaking the blocks and chunksize
> parameters.
>
> Good luck!
> Mike Polek
> Pictage, Inc.
Thank you, but the results are the same. Another idea?

Ralf.