NFS problems with recent -current

From: Peter Jeremy <peterjeremy_at_optushome.com.au>
Date: Sat, 23 Jun 2007 08:24:06 +1000
I have 7.0-CURRENT/i386 from 7th June running as an NFS server with
a 6-STABLE/amd64 client and the server is regularly getting flooded
with the following:
Jun 23 04:00:45 server kernel: uma_zalloc_arg: zone "mbuf" with the following non-sleepable locks held:
Jun 23 04:00:45 server kernel: exclusive sleep mutex nfsd_mtx r = 0 (0xc07ec040) locked _at_ /usr/src/sys/nfsserver/nfs_srvsock.c:660
Jun 23 04:00:45 server kernel: KDB: stack backtrace:
Jun 23 04:00:45 server kernel: db_trace_self_wrapper(c071c52b,d630ab10,c0585bcf,c071c8af,d630ab24,...) at db_trace_self_wrapper+0x26
Jun 23 04:00:45 server kernel: kdb_backtrace(c071c8af,d630ab24,4,1,0,...) at kdb_backtrace+0x29
Jun 23 04:00:45 server kernel: witness_warn(5,0,c073093c,c0721815,d630ab34,...) at witness_warn+0x1bf
Jun 23 04:00:45 server kernel: uma_zalloc_arg(c1044000,d630ab88,2,8,c3147d24,...) at uma_zalloc_arg+0x34
Jun 23 04:00:45 server kernel: nfs_realign(c07ec040,0,c072c0f7,294,0,...) at nfs_realign+0x6f
Jun 23 04:00:45 server kernel: nfsrv_rcv(c3103dec,c3147d00,2,168,0,...) at nfsrv_rcv+0x49a
Jun 23 04:00:45 server kernel: nfssvc(c3fa1c00,d630acfc,8,c,c075df08,...) at nfssvc+0x77f
Jun 23 04:00:45 server kernel: syscall(d630ad38) at syscall+0x29a
Jun 23 04:00:45 server kernel: Xint0x80_syscall() at Xint0x80_syscall+0x20
Jun 23 04:00:45 server kernel: --- syscall (155, FreeBSD ELF32, nfssvc), eip = 0x280c9d17, esp = 0xbfbfe8bc, ebp = 0xbfbfe8d8 ---

The server seems to run for some time and then start generating these
errors.  I seem to be able to stop them by unmounting the client and
restarting the NFS server processes (though once the server
deadlocked).

So far, I've got records of it starting:
Jun 11 or Jun 12 (I've lost the first error in this block)
Jun 15 22:15:01
Jun 21 22:36:00
Jun 22 22:59:56

I'm not sure what the underlying trigger is - I'm not around at those
times (though I do leave mutt running on the client).  The times don't
match any periodic jobs I've got running and don't match
fetchmail/sendmail activity on the server.

Has anyone else seen this or have any ideas where to look?

-- 
Peter Jeremy

Received on Fri Jun 22 2007 - 20:24:09 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:13 UTC