[05:30:55] --- squinney has become available [06:23:41] --- meffie has become available [06:35:48] --- meffie has left [06:37:06] --- meffie has become available [06:44:22] --- stephan.wiesand has become available [06:45:37] test test [06:53:29] --- Marc Dionne has become available [07:01:12] Hello All [07:01:23] hi stephan [07:01:27] --- deason has become available [07:01:37] --- ktdreyer has become available [07:01:59] morning / afternoon all [07:02:21] Greetings from LinuxTag in Berlin. [07:02:51] we should have a booth here, or a presentation... [07:03:42] Let's start with the bug reported with Linux 3.9.2 [07:03:55] Marc, Andrew, what' your stance? [07:04:17] i can provide a minimal fix to make tmpfs work [07:04:36] for btrfs, i did some testing but did not see any issues [07:04:52] I was a bit surprised that tmpfs is considered supported at all? [07:05:32] --- kaduk@mit.edu/barnowl has become available [07:05:33] define "supported".. :) it happens to work with the API we use, at least until this change [07:05:58] but not sure that it's widely used [07:06:40] my "stance": well, I'm fine with it if the fix is small :) [07:06:46] I just haven't seen it yet [07:07:23] i'll push it in a minute so you can have a look [07:07:29] If the change making tmpfs work is simple, why not. I'm not sure it would qualify as the only reason for another prerelease though. [07:09:09] I could try btrfs for a bit, see what happens [07:09:24] does fedora 18 have the kernel he mentioned? [07:09:30] 9928 [07:09:48] yes, fedora is currently at 3.9.2 [07:10:33] but that will change within days... [07:12:27] 9928 doesn't look exactly trivial to me [07:13:02] it doesn't look that bad to me, but it could be more trivial if we didn't cache those results [07:13:43] --- stephan.wiesand has left [07:14:28] --- stephan.wiesand has become available [07:15:00] --- Simon Wilkinson has become available [07:15:04] so are all of these code paths just for readahead? does that mean we can't copy the pages in advance for readahead stuff? [07:15:11] er, for tmpfs [07:15:54] well it effectively disables readpages(), and the "fastpath" case [07:18:06] Ok, once it was accepted on master, please pull it up for 1.6. If not in 1.6.3, we'll want it in 1.6.4. [07:18:20] yeah, it just seems odd to me if we _can't_ populate the pages in advance like normal, even if the data is there [07:19:34] Let's go on to another potential reason for having a pre3: Linux 3.10 [07:19:37] er, but I don't mean to discuss details of that in here; we can move on [07:19:41] yes [07:19:54] If you can't use readpage, you can't populate the pages in the background [07:20:04] Marc, are the two patches for master all that's required? [07:20:26] yes, so far those 2 patches are all that's needed for the current mainline [07:21:01] also had to get a patch in the kernel to un-GPL some stuff, but that's not in our code.. [07:21:47] more GPLONLY fun? [07:22:27] When will 3.10 be released? [07:22:58] yes, schedule_work() went GPL only as a side-effect of some function inlining - but I go t a ptach in to un-GPL the symbols that are needed [07:23:46] it's early in the cycle, so probably another 6-7 weeks before release [07:25:02] Those two look harmless to me. [07:25:43] It would be nice to have them available for 1.6 too, like the tmpfs fix. [07:26:36] It would be good to have 3.10 support in 1.6.3 [07:26:43] We don't have to decide today whether we'd do a pre3 just for those Linux changes. But opinions are most welcome. [07:27:30] i'll push the 3.10 fixes to 1.6 once they are merged in master. for the tmps fix, i'll push a version of it to master. [07:28:01] If we have another reason for a pre3, the 3.10 changes should really go in. [07:28:02] so far I don't know if they need another pre, but I'd include them without blocking on them [07:28:59] You mean, if no further problems are reported, add them and release w/o public testing? [07:30:08] But again, no need to decide today. [07:30:29] Marc: Thanks. [07:30:39] Lets talk about binaries? [07:31:37] Stephen, I'm receiving the notification mails now, so I know you started building for RH. Thanks. [07:31:39] I'd be okay with that for the tmpfs change if it were simpler; maybe the other ones if they had other specific testing [07:32:06] Andrew: noted. [07:32:40] I'll upload the RH binaries as I find the time, hopefully starting today. [07:32:52] But what else do we want? [07:34:20] Derrick, for 1.6.1 you just uploaded an OSX installer - you'd still do that now, right? [07:35:39] I figure Christof will build for SuSE. He did for pre1. [07:36:35] I think I need to free up some disk space on my build machine before I can build for freebsd. It's just copying off the tarballs of old build chroots, but there seem to be too many things to do... [07:37:30] Whatever you get done, I'll upload. [07:38:12] It would be nice to have some advance notice if someone even plans to build anything. [07:38:30] I've not had a chance to sort out the EL5 packages yet [07:38:52] Everything else appears to be building fine. Will try to test the EL6 packages fairly soon. [07:39:36] I know the source builds on EL5, and the spec didn't change. [07:41:32] If it doesn't build, file a bug ;-) [07:42:16] NB we have test packages in SL/6rolling/testing, announced yesterday. No feedback yet either... [07:42:50] Anything else we should discuss today? [07:43:57] well, what's our schedule for pre2? [07:44:14] how long are we waiting for feebdack, absent any other significant developments? [07:45:09] good question. it would be easier to answer if there were any feedback. [07:46:16] but I think, at the very least two weeks after the announcement [07:47:34] I'll send mail to openafs-info once we have a few binaries up there, which should be soon. Maybe it will help. [07:48:39] Are there any pressing reasons to get 1.6.3 out a.s.a.p.? [07:50:10] i'm not aware of any. [07:50:10] I guess not. [07:50:50] If nothing else happens, we should start planning for 1.6.4 next week :-) [07:50:54] let's see 1.6.2 was out on march 4, so about 3 mouths ago. [07:51:24] so, this seems like a good pace to me. [07:53:23] I think it's not too bad. [07:55:09] I'd guess that >95% of systems are still running pre-1.6.2. [07:56:02] Some of you have large clients/deployments - what are they doing? [07:57:22] well, the larger sites take way longer than that to move [07:57:32] well, it depends [07:58:06] We (for a value of "we") are sticking to what's in ubuntu precise, which is 1.6.1-1+ubuntu-mumble [07:58:20] okay, it depends, but at least some are still just evaluating/investigating 1.6, seeing what release they want to pick and what patches to add [07:59:57] but a lot of the patches they run are from 1.6 releases, at least [08:00:41] sadly, i sense many sites are still on 1.4.x series, but having a steady release cycle for 1.6 i think will help immensely. [08:01:11] My production servers are 1.4 :-) [08:01:41] But that will start changing next week. [08:01:42] for example :) [08:02:25] sorry. yes, i will deliver you some macos installers [08:02:35] But I agree with your opinion on the value of steady releases. [08:02:50] Derrick: Thanks ;-) [08:03:25] My production servers are 1.4.8, but they're up for an OS upgrade, uh, last week. I expect to take tip-of-1.4 at that time, but 1.6 in a month or two. [08:04:15] It's too early to set a deadline. But my feeling is that we should tag 1.6.3 within "a few weeks". [08:04:31] Unless there;s a serious bug reported, pf course. [08:04:42] Right. [08:04:56] We've got about 1000 machines moving to 1.6.2 next week [08:05:12] from 1.4.x? [08:05:21] no, we're on 1.6.1 [08:05:41] 1.6.2 + a few patches, I guess? [08:05:52] Once that's done I will have more chance to test 1.6.3 [08:06:20] we only have a couple of 1.4.14 clients hanging around for various odd reasons [08:06:30] kaduk: I don't know if the tip of 1.4.x has really been tested... 1.4.14.1 I think would be a safer bet [08:07:11] "tip of released 1.4" [08:07:23] okay [08:07:57] Let's face it: testing in the field happens *after* the release. [08:08:15] So let's release 1.6.3 rather soon. [08:08:51] Anything else for today? [08:10:25] It seems not. So I'll go meet the other geeks ;-) [08:10:38] Thanks everyone. [08:10:57] --- stephan.wiesand has left [08:11:16] --- meffie has left [08:47:14] --- Marc Dionne has left [09:23:59] --- squinney has left [09:40:48] --- Simon Wilkinson has left [09:47:41] --- Derrick Brashear has left: Disconnected [14:20:14] --- Derrick Brashear has become available [14:32:37] --- Derrick Brashear has left: Disconnected [14:37:23] --- Simon Wilkinson has become available [15:01:07] --- Derrick Brashear has become available [15:13:49] --- Simon Wilkinson has left [15:15:33] --- Simon Wilkinson has become available [16:05:48] --- Derrick Brashear has left: Replaced by new connection [16:05:56] --- Derrick Brashear has become available [16:35:28] --- Derrick Brashear has left: Disconnected [16:36:10] --- Derrick Brashear has become available [16:44:17] --- deason has left [18:13:54] --- Derrick Brashear has left: Disconnected [22:55:22] --- Derrick Brashear has become available [23:33:33] --- Simon Wilkinson has left [23:42:42] --- Simon Wilkinson has become available [23:42:55] --- Simon Wilkinson has left [23:52:52] --- Simon Wilkinson has become available [23:55:09] --- Simon Wilkinson has left