[ewg] OFED 1.2 April 16 meeting summary

[Tziporet Koren]

OFED 1.2 April 16 meeting summary 
 
Main decisions:
1. RC2 will be ready on the US morning of Wed April 18. (RC2 date is
derived from Intel schedule for the 256 nodes cluster)
2. RC3 due date is April 26. 
3. All release notes and other documents should be ready for RC3
4. Bug fixes after RC3 will have to be approved by the RM (Tziporet)
5. Open MPI version will be replaced in RC3. This will bring the vanilla
Open MPI version to OFED.
6. Official release will be done after the Sonoma conference.
 
Cluster testing:
- Intel will test IPoIB, Intel MPI and MVAPICH on 256 nodes cluster.
- The labs will test Open MPI and MVAPICH on 256 nodes cluster
 
Note: Next coordination meeting will be on April 25,  Wed 11:30am PST
 
Bugs review (bugs are updated in bugzilla):

bug_id	 bug_severity	 assigned_to	 short_short_desc	
420	 critical	 monil at voltaire.com <mailto:monil at voltaire.com>
PKey table reordering caused by SM failover stops ipoib traffic  - will
not make it to RC2 	
513	 critical	 rjwalsh at pathscale.com
<mailto:rjwalsh at pathscale.com> 	 error while installing ipath driver  -
Qlogic should decide today if they want to support this platform and OS	
		<mailto:e at opengridcomputing.com> 		
534	 critical	 vlad at mellanox.co.il
<mailto:vlad at mellanox.co.il> 	 SLES9 - Installer fails on declarations
- OFED 1.2-20070409  - need to check if this kernel version is provided
by Novell 	
465	 critical	 mst at mellanox.co.il <mailto:mst at mellanox.co.il>
IPoIB CM HA fails after several hours of failovers  - Michael will try
to reproduce 	
539	 critical	 tziporet at mellanox.co.il
<mailto:tziporet at mellanox.co.il> 	 "Catastrophic error detected"
while running IPoIB bonding port failover test  - Scott to test FW
version with a fix 	
543	 major	 eitan at mellanox.co.il <mailto:eitan at mellanox.co.il>
ibis fails to compile on OFED-1.2-20070415-0141  - fixed 	
499	 major	 vlad at mellanox.co.il <mailto:vlad at mellanox.co.il>
module compiled over ofed won't load due to symbol version mismatch  -
Vlad and Moni to agree on solution 	
459	 major	 monis at voltaire.com <mailto:monis at voltaire.com>
support ib-bonding on RHEL4U4/RHEL5, put kernel name in RPM name, and
clean up better  - fixed in the new bonding package 	
508	 major	 mst at mellanox.co.il <mailto:mst at mellanox.co.il>
IPoIB CM multicast is hogging interrupts	
506	 major	 mst at mellanox.co.il <mailto:mst at mellanox.co.il>
IPoIB IPv4 multicast throughput is poor  - Tziporet to provide a FW
version that should fix the issue 	
527	 major	 monis at voltaire.com <mailto:monis at voltaire.com>
ib-bonding won't compile for RHEL5 i686 PAE kernel  - fixed in the new
bonding package 	
538	 major	 monis at voltaire.com <mailto:monis at voltaire.com>
integrate IPoIB bonding with IPoIB CM- fixed in the new bonding package	
541	 major	 monis at voltaire.com <mailto:monis at voltaire.com>
slow failover with IPoIB bonding  - need investigation 	
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/ewg/attachments/20070416/d6f27eef/attachment.html>