kernel BUG at net/sunrpc/sched.c:902 on server-state-recovery and nfs-server-stable git branches

J. Bruce Fields bfields at fieldses.org
Thu Jan 25 16:12:55 EST 2007


On Thu, Jan 25, 2007 at 01:01:31PM -0800, Bryce Harrington wrote:
> On 1/25/07, Bryce Harrington <bryceharrington at gmail.com> wrote:
> > The following bug is present in the git HEAD of the
> > server-state-recovery branch (from Jan 16th).  The bug also occurred
> > on the nfs-server-stable branch Jan 6th and 9th, but has not been
> > present in the 3 snapshots taken since then.
> >
> > The bug occurred at the start of testing during the pynfs run.
> 
> This may also be of interest.  This was in the console log on the server:
> 
> http://crucible.osdl.org/runs/3635/sysinfo/nfs12.console
> 
> end_request: I/O error, dev hda, sector 0
> NFSD: setclientid: string in use by client(clientid 45b69ace/00000002)
> xdr error! (fs/nfsd/nfs4xdr.c:1165)
> NFSD: preprocess_seqid_op: bad seqid (expected 1, got 50)
> NFSD: preprocess_seqid_op: magic stateid!
> xdr error! (fs/nfsd/nfs4xdr.c:316)
> xdr error! (fs/nfsd/nfs4xdr.c:413)
> xdr error! (fs/nfsd/nfs4xdr.c:413)

This is all fairly normal if you're running the pynfs tests.

> nfs4_cb: server client1_pid19094x86-64.so.2 not responding, timed out
> nfs4_cb: server client1_pid19094x86-64.so.2 not responding, timed out
> nfs4_cb: server Badid_for_CID2_pid=19094 not responding, timed out
> nfs4_cb: server Clientid_for_CID3_pid=19094 not responding, timed out
> nfs4_cb: server Clientid_for_CID4_pid=19094 not responding, timed out
> nfs4_cb: server Clientid_for_CIDCF2_pid=19094 not responding, timed out
> nfs4_cb: server Clientid_for_CIDCF3_pid=19094 not responding, timed out
> nfs4_cb: server Clientid_for_CIDCF3_pid=19094 not responding, timed out
> nfs4_cb: server Clientid_for_CIDCF3_pid=19094 not responding, timed out
> nfs4_cb: server client1_pid19094x86-64.so.2 not responding, timed out

This I'm not sure about; could be the client crashed and isn't
responding any more, though I'm a little surprised the server didn't
give up sooner.  Probably safe to ignore it for now in any case.

--b.


More information about the NFSv4 mailing list