Difference between revisions of "LWG Minutes 2015-06-03"
(Created page with "== Agenda == * Lustre 2.8 update * Upstream client update * lustre.org update == Attendance == * Chris Morrone (LLNL) * Cory Spitz (Cray) * Chris Horn (Cray) * Richard Henw...")
Revision as of 10:48, 23 June 2015
- Lustre 2.8 update
- Upstream client update
- lustre.org update
- Chris Morrone (LLNL)
- Cory Spitz (Cray)
- Chris Horn (Cray)
- Richard Henwood (Intel)
- Chris Hanna (IU)
- Justin Miller (Cray)
- Peter Jones (Intel)
- Andreas (Intel)
Chris to send reminder to lustre-devel
SLES 12 server - James - ldiskfs work needed Peter - OS support we usually allow to land up to Code Freeze date
Upstream client update
James - Making pretty good progress. Some more people starting to step up and contribute like Purdue guys. Oleg made sysfs support. Good for kernel, but our tools all depend on proc, so that is now an issue.
Andreas - are we ready to move it kernel devel contact to be email@example.com list?
Message hold without warning email is desirable.
James also starting to backport patches from upstream kernel client to external lustre repo.
Upstream hates when strings broken up. Would rather go over 80 characters than break strings. Lustre checkpatch script needs update to complain less about known exceptions like that. Upstream they apparently also allow comments for structure members to exceed 80 characters.
Aligning variables declared at top of function is not prefered in upstream kernel.
Need to update Lustre wiki about these style changes. Peter suggested moving those docs to wiki.lustre.org when they need to be updated, and then change hpdd wiki to just have pointers to wiki.lustre.org instead of the content.
Cory - email out about coordinating effort...James can you send email out?
James - Sure, and we should maybe set up chat channel of some kind.
Andreas pointed out that we have have an IRC lustre channel on freenode, and he is the owner/moderator for the channel.
Justin to update wiki with IRC channel details.
Cray looking to get Ben Evans involved with client upstreaming effort.
James - We need to keep the client upstreaming work going after this crisis is over (so we don't wind up in another crisis again in the near future).
Quick start Scripting and explanation of setting up virtual clusters
"Lustre Monitoring and Statistics Guide" from lustre.org
Andreas - if you see something from wiki.old.lustre.org pop up in a google search, that is a good candidate to more to the new wiki.lustre.org.