[OpenAFS] afs-clien on afs-server hangs ... (again !)

[Baus, Hans-Joerg]

with /var/log/messages:

Jan 12 01:03:32 geeko kernel: Increase -stat parameter of afsd(VLRU
cycle?)<1>Unable to handle kernel paging request at virtual address
ffffffff
Jan 12 01:03:32 geeko kernel:  printing eip:
Jan 12 01:03:32 geeko kernel: e08e75b1
Jan 12 01:03:32 geeko kernel: *pde = 00001063
Jan 12 01:03:32 geeko kernel: *pte = 00000000
Jan 12 01:03:32 geeko kernel: Oops: 0002
Jan 12 01:03:32 geeko kernel: CPU:    0
Jan 12 01:03:32 geeko kernel: EIP:    0010:[<e08e75b1>]    Not tainted
Jan 12 01:03:32 geeko kernel: EFLAGS: 00010286
Jan 12 01:03:32 geeko kernel: eax: 0000002d   ebx: e090e550   ecx:
dcf6c000   edx: 00000000
Jan 12 01:03:32 geeko kernel: esi: cec9dd24   edi: dcf6de04   ebp:
00007d02   esp: dcf6dccc
Jan 12 01:03:32 geeko kernel: ds: 0018   es: 0018   ss: 0018
Jan 12 01:03:32 geeko kernel: Process save (pid: 28552,
stackpage=dcf6d000)
Jan 12 01:03:32 geeko kernel: Stack: e08ffe60 e08c71ee 00000010 00000246

ffffff0b e08c7246 00000010 e08c721d
Jan 12 01:03:32 geeko kernel:        e08ffe60 e08c71ee 00000010 00000246

e096ca00 00000000 dcf6de04 dcf6c000
Jan 12 01:03:32 geeko kernel:        e131e3c0 cec9dc00 00000005 e08d0b74

dcf6de04 00000000 00000000 e08d0be4
Jan 12 01:03:32 geeko kernel: Call Trace:    [<e08ffe60>] [<e08c71ee>]
[<e08c7246>] [<e08c721d>] [<e08ffe60>]
Jan 12 01:03:32 geeko kernel:   [<e08c71ee>] [<e08d0b74>] [<e08d0be4>]
[<e08bd205>] [<e08bd249>] [<e08bcf57>]
Jan 12 01:03:32 geeko kernel:   [<e08d251c>] [<e08d258d>] [<e08efbef>]
[<e08f2d09>] [<c0135e8b>] [<c01365b7>]
Jan 12 01:03:32 geeko kernel:   [<c0135c0d>] [<c0136856>] [<c0136be1>]
[<c0133d4d>] [<c010867b>]
Jan 12 01:03:32 geeko kernel:
Jan 12 01:03:32 geeko kernel: Code: c6 05 ff ff ff ff 2a 83 c4 10 83 c4
0c c3 90 55 57 56 53 8b


I have had the same problem a few months ago and increased the -stat
parameter (thanks to Derek Atkins!)to

OPTIONS="-stat 16000 -dcache 4000 -daemons 6 -volumes 256 -files 50000"

and the cache (/usr/vice/etc/cacheinfo) to

/afs:/usr/vice/cache:200000


Now the problem is back - and i changed the parameters now to

OPTIONS="-stat 32000 -dcache 6000 -daemons 6 -volumes 256 -files 50000"

/afs:/usr/vice/cache:300000

and i guess it may work for a few months more... but i guess also that
our networker backup software is somehow involved with the problem and
the amount of data grows further and the problem comes back again...


Has anyone had the some experience (with legato networker)?

Could an upgrade to 1.2.8 help (my System is linux-2.4.19-xfs)?

Thank You All For Your Help!


-- Jörg

Dipl.-Ing. Jörg Baus
Krauss-Maffei Wegmann GmbH&Co.KG
Abteilung W-AS 22
Systemmanagement

Hausadresse:                 Postfachadresse:
Krauss-Maffei-Str. 11        Postfach 500240
80997 München                80972 München

Tel.:  ++49 (89) 8140-4025
Fax:   ++49 (89) 8140-4912
Email: joerg.baus@kmweg.de