- Wednesday, March 25, 2015

release-team@conference.openafs.org

Wednesday, March 25, 2015< ^ >

Room Configuration

Room Occupants

GMT+0
[09:00:06] Jeffrey Altman leaves the room
[09:01:41] Jeffrey Altman joins the room
[13:05:23] meffie joins the room
[13:51:03] <Jeffrey Altman> I will be unable to participate in today's meeting.
[13:51:31] wiesand joins the room
[13:54:18] Marc Dionne joins the room
[13:57:59] <wiesand> Jeffrey: Bad new, but Thanks.
[13:58:08] <wiesand> Marc: Hi! Glad you’re here :)
[13:58:19] <Marc Dionne> howdy
[13:58:56] <wiesand> Have you had time to loo at 4.0-rc5 yet?
[13:59:25] <wiesand> I hear you were having some snow related troubles
[13:59:32] <Marc Dionne> i run mainline on one of my systems (updated daily), usually with YFS.  doing a build of openafs now and it looks ok
[14:00:00] <wiesand> Good. Do we expect an rc6 or rc7?
[14:00:01] <Marc Dionne> yeah early last week we had 100+ cm in 3 days.  caused a bit of problems.. :)
[14:00:16] <wiesand> Wow.
[14:00:35] <Marc Dionne> i would expect at least another rc
[14:01:32] <wiesand> But not necessarily more than one, right?
[14:02:26] <Marc Dionne> no, looks like things have been fairly quiet
[14:02:58] <wiesand> I’m thinking about changing plans.
[14:03:12] <shadow@gmail.com/barnowlE5B64A04> to?
[14:03:22] <Marc Dionne> not many upcoming changes in linux-next either (for following release), but that doesn't necessarily mean much
[14:03:35] <wiesand> Make a 1.6.12.1 with linux 4.0 changes, and probably ben’s FBSD build fix.
[14:03:44] <shadow@gmail.com/barnowlE5B64A04> ok
[14:03:47] <wiesand> Leave the ubik changes for a later release.
[14:04:12] <wiesand> Because I think we should have more than a week of pre1 before pushing out the ubik stuff.
[14:04:28] <wiesand> But Linux 4.0 is close.
[14:04:33] <wiesand> Does this make sense?
[14:05:53] <wiesand> We haven’t merged anything since the 1.6.11 release, so we don’t even need a branch.
[14:07:55] <wiesand> Ok. Unless there are objections raised after I send the minutes to release-team, that’s what we’ll do.
[14:08:33] <wiesand> I’m wondering about gerrit 11809...
[14:09:07] <wiesand> It doesn’t break the build on FBSD 9.3, but is the a chance that the module will no longer work without that flag on older releases?
[14:09:39] <Marc Dionne> the commit message suggests it may have been a noop in the past
[14:10:19] <wiesand> Right, and if we go for 1.6.12.1 worst case is 10.1 need that and older releases need 1.6.11.
[14:10:56] <wiesand> So, I’m much inclined to merge that soon just to get automatic buildbot verifications back.
[14:11:34] <wiesand> Marc: not for the very next release, but did you see my reply to the RT ticket?
[14:14:04] <Marc Dionne> what's the number again
[14:14:18] <wiesand> 132020
[14:15:57] <Marc Dionne> yeah not sure why you're also seeing it with those other sizes.  but a) it's a very long standing problem.  b) i experimented with some simpler workarounds.  they work, but just leave other different corner cases
[14:16:28] <wiesand> :-(
[14:16:48] <wiesand> When a system has this problem, will it help to change the cache size on the fly?
[14:17:05] <wiesand> Yes, I could simply try.
[14:17:09] <Marc Dionne> my first idea was not to do any writes if you have only 1 chunk.  but then what if you're writing a lot of small (<1 chunk) files at the same time...
[14:17:51] <Marc Dionne> i would say more likely with a smaller cache, and also more likely with larger chunk sizes
[14:19:17] <wiesand> Probably. But changing the cache size a few dozen chunksizes up or down would be doable without interrupting work on the system.
[14:19:46] <Marc Dionne> that needs a client restart, no?
[14:19:59] <wiesand> fs setcache? I think not?
[14:21:22] <Marc Dionne> hmm, but does that change the number of chunks in play.  need to read a bit of code
[14:22:04] <wiesand> That’s what I was wondering about. I’ll try it, at the latest when my users have that problem again ;-)
[14:22:05] <Marc Dionne> what i want to try is that if you find you have a single chunk to play with, and you're doing a partial write, you should write 1 chunk from a different file
[14:23:07] <wiesand> I’m willing to test any patches you might have. But frankly I don’t have much time on my hands these days.
[14:23:08] <Marc Dionne> then if you keep writing to the same file, next time you get there you should have 2 chunks in play
[14:24:06] <Marc Dionne> well not a high priority item for me either, given it's a very old bug.  but would be nice to get it fixed
[14:25:26] <wiesand> Oh yes.
[14:25:43] <wiesand> I’ll just keep it on the agenda ;-)
[14:26:10] <wiesand> I guess there’s not much more to discuss today?
[14:26:37] <wiesand> I forgot: do we issue prereleases for those point releases like 1.6.12.1?
[14:27:01] <meffie> i do not think so.
[14:27:44] <wiesand> Fine :)
[14:27:53] <meffie> looking at the tags, i dont see any
[14:27:55] <Marc Dionne> 1.6.11.1?
[14:28:07] <wiesand> Er, sure. Sorry.
[14:28:54] <wiesand> Mike, you just made it onto the participants list ;-)
[14:29:16] <meffie> sorry, just been listening.
[14:29:25] <wiesand> Anyway, anything else to discuss today?
[14:29:34] <meffie> no.
[14:30:10] <wiesand> Fine. Thanks a lot everyone!
[14:30:35] <wiesand> I’ll send minutes later. Bye.
[14:30:39] wiesand leaves the room
[14:32:28] Marc Dionne leaves the room
[14:43:08] meffie leaves the room
[18:35:21] kaduk joins the room
[19:30:07] kaduk leaves the room
[19:33:50] shadow@gmail.com/barnowlE5B64A04 leaves the room
[19:34:23] shadow@gmail.com/barnowlE5B64A04 joins the room
[21:51:35] kaduk joins the room
[22:06:26] kaduk leaves the room
[22:06:26] kaduk joins the room