From: Dr J Pelan (J.Pelan@gatsby.ucl.ac.uk)
Date: Sat May 24 2003 - 13:42:19 EDT
OS: Tru64 v5.1B (patched with T64V51BB22AS0002-200 Tru64 V5.1b ECO)
H/W: DS20 & DEGPA-TX x2
I have a 2 node TruCluster server which has crashed with a kernel panic
originating in the NFS v3 *server* code. All the NFS clients are running
Linux (2.2 & 2.4 kernels) and the current *assumption* is that one or more
of these are sending some sort of malformed NFS request which Tru64 can't
handle robustly. Obviously, this will have to be escalated with HP,
support contract permitting, but list members may like to be aware of the
issue.
Notes;
o The system crashes with/without the latest ECO patches.
(Patch 872 looks like it may address this problem but no go).
o The behaviour is very recent (perhaps related to newer Linux kernels ?)
o No tcpdumps *yet* so cannot determine which clients are involved and
cannot reproduce it. It has now occurred three times, circa every 6 days.
o This is a very effective Denial of Service as the cluster is taken out.
o This must be a common configuration - has it been seen elsewhere and
if not why not ?
o Crash dumps etc. are available.
-- John P.
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:20 EDT