Here are the high level minutes from today's Lustre
2.1 call
Attendees (apologies if I missed anyone)
Bull: Sebastien Buissson
Cray: Wally Wang, Cory Spitz
DDN: David Vasil
LLNL: Pam Hamilton, Chris Morrone, Prakash
OpenSFS: Shay Seager
Oracle: Kevin van Maren
ORNL: (no attendees)
PNL: (no attendees)
PSC: (no attendees)
SGI: (no attendees)
TACC: John Hammond
Terascala: (no attendees)
Whamcloud: Peter Jones, Oleg Drokin, Richard
Henwood, Sarah Liu
Xyratex: Vitaly Fertman
Review of actions from last meeting
======================
ACTION 2011-03-01 Xyratex to explore whether any
existing Xyratex 2.x patches are appropriate for
inclusion in the 2.1 release
-No progress on testing of LU145 ONGOING
ACTION 2011-03-01 ALL to contribute to 2.1 testing
-No news this week ONGOING
ACTION 2011-04-26 LLNL to advise when RHEL6 testing
is available on Hyperion
-We have a path forward but Hyperion testing focused
on 1.8.x atm ONGOING
ACTION 2011-06-14 Jones to check whether asymmetric
router failure detection can be turned on by default
for 2.1
-According to the information in
https://bugzilla.lustre.org/show_bug.cgi?id=23575#c60
this would not be appropriate COMPLETE
Testing update
==============
-TACC 2 software RAID ; Performance seems far worse;
John will open a ticket soon; van Maren warned of
24264 data corruption issues for anyone using
software RAID
-ORNL benchmarking with Cray Gemini system later
this week; James uploaded SLES support patches into
gerrit under LU355 and looking for testing feedback
-LLNL Early in testing cycle; will be doing larger
scale tests on latest code
-Cray installing CentOS this week so should be able
to use this in testing going forward
-Terascala no information
-Whamcloud found solution for LU387 which had been
regularly affecting MMP tests in automated
regression runs
Blocker update
==============
-LU388 Enables OFED testing on Toro; should be fixed
ahead of next tag
-LU396 LLNL hit once during RHEL6.1 scale testing;
will drop as a blocker
-LU394 Patch going through inspections and testing
-LU437 LLNL hit regularly during RHEL6.1 scale
testing; engineer investigating
-LU442 Hit in production at CEA; locking needs to be
reworked
-LU386 Fuller logs required; should gather these
today
AOB
===
-Xyratex have found a data corruption issue; Jones
expressed concern that this information had not been
shared more widely given that some sites are running
2.x code in production; Fertman said that a JIRA
ticket would be opened by EOB today
ACTION 2011-06-21 Xyratex to open JIRA ticket with
details of data corruption issue
Next meeting will be at 9:30 AM PT on June 28th.
Please send suggestions for agenda topics ahead of
that time.
Regards
Peter