[pnfs] NFS4ERR_BADSESSION in connectathon testing

[William A. (Andy) Adamson]

Hi

I'm testing my Austin final patch set, which includes setting the clientid
in the LOCK/LOCKT and LOCKU operations to zero. The tests pass - sometimes!
When they fail, it is usually in the connectathon lock tests, after I've run
either the basic or special tests. The lock tests have (so far) always run
successfully when run first. The lock tests always fail on test #12, the 2nd
time through test series.

The error is on the client which sends two compounds on the wire with the
same session sequence number and slot number which causes the server to
correctly return NFS4ERR_BADSESSION, and the client hangs.

Connectathon lock test #12 involves a child and parent one holds the lock,
the other waiting, and is then signaled. So there is a timing piece to the
test.

Has anybody else seen this behaviour? I don't think it is due to the LOCK
patches, and I don't think anyone else ran the Connectathon lock tests in
Austin - is that right?

-->Andy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://linux-nfs.org/pipermail/pnfs/attachments/20070703/9981fc5e/attachment.html