Test results for linux-2.6.18-rc6-CITI_NFS4_ALL-1

[Bryce Harrington]

On Mon, Sep 11, 2006 at 12:45:00AM -0400, J. Bruce Fields wrote:
> http://www.citi.umich.edu/projects/nfsv4/linux/kernel-patches/2.6.18-rc6-1/linux-2.6.18-rc6-CITI_NFS4_ALL-1.diff
> 
> Changes since 2.6.18-rc5-CITI_NFS4_ALL-1
> 	- update to -rc6 and Trond's latest
> 	- also include all nfsd patches from -mm
> 	- nfsd4 acls: minor cleanup
> 	- nfsd4 acls: reject some explicit denies that are unenforceable
> 	- aglo: clean up spkm3 update based on comments from trond
> 	- gss: remove print_hexl's, minor style cleanup
> 
> The patches in -mm (mainly from Neil Brown, Greg Banks, and Olaf Kirch) include
> a ton of stuff; see
> http://www.linux-nfs.org/cgi-bin/gitweb.cgi?p=bfields-2.6.git;a=summary
> for details.
> 
> I might put out another version or two over the next week just to collect any
> connectathon bug fixes.

Hi Bruce,

We've now combined the CITI releases into the git tree branch report, so
there'll be a single place to go for both info, and to make it easier to
compare with the git trees:

    http://crucible.osdl.org/runs/branches.html

This one is still seeing some issues, that may be fixed on some of the
branches:

   * newpynfs skipping tests "Dependency MKFILE st_open.testOpen had
     status FAILURE."  We had seen this commonly on the
     nfs-client-stable branch, but we found it went away after applying
     David Howells' fs_cache patch to Trond's tree.

   * Connectathon failing on './test6: readdir' and skipping a bunch of
     tests after.  I think this may be a new issue.

You'll also notice ffsb is showing much lower transactions per second on
x86 than normal (469 vs. ~1176 in prior runs of nfs-server-stable).  I
am going to rerun the test again to see if it is a recreatable issue.
I don't know that much about ffsb yet, so am curious to see if it's a
legit issue, or just a test config problem.

Bryce