Red Hat Bugzilla – Bug 70118
NFS kernel module crashes on high write load
Last modified: 2007-04-18 12:44:52 EDT
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; .NET CLR
Description of problem:
This seemed at first similar to bugs 64921, 64984 and 65772, as fixed in
2.4.18-5, but we are still experiencing kernel panic problems under high write
load of the nfs server.
All clients mount with auto,noac,async,rsize
kern.log shows the following:
Jul 30 00:40:31 lifetimefs1 kernel: Assertion failure in
ion() at commit.c:535: "buffer_jdirty(bh)"
Jul 30 00:40:31 lifetimefs1 kernel: ------------[ cut here ]------------
Jul 30 00:40:31 lifetimefs1 kernel: kernel BUG at commit.c:535!
Jul 30 00:40:31 lifetimefs1 kernel: invalid operand: 0000
Jul 30 00:40:31 lifetimefs1 kernel: autofs nfs nfsd lockd sunrpc eepro100
ipchains usb-ohci usbcore ext3 jbd cciss
At this point all is dead, even console logins don't work and a powercycle is
the only solution.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. High write load on the NFS server
Actual Results: kernel panic and server is doa!
Expected Results: NFS server keeps running :-)
Is it always the same crash?
99% or more of the occurrences of this problem were fixed in 2.4.18-4smp, but I
continued to get just a few rare reports even with the fixed kernel, and found
another possible cause of the same problem since then. That is fixed in our
internal tree and will be part of the next errata kernel.
for a 2.4.18-7 kernel which should hopefully fix this problem.
*** This bug has been marked as a duplicate of 68026 ***