- Wednesday, January 11, 2017

release-team@conference.openafs.org

Wednesday, January 11, 2017< ^ >

Room Configuration

Room Occupants

GMT+0
[11:30:19] mvita leaves the room
[14:28:27] jgorse leaves the room
[14:28:27] jgorse joins the room
[14:28:35] jgorse leaves the room
[14:28:50] mvita joins the room
[14:49:32] wiesand joins the room
[14:53:59] jgorse joins the room
[14:57:14] <kadukoafs@gmail.com/barnowlE45450E9> Sorry, I won't be able to make the meeting; I'm under the weather.
[14:57:35] <jgorse> Hope you get well soon, Ben.
[14:58:46] <jgorse> And I hope to know what I can do to help merge https://gerrit.openafs.org/#/c/11794 into master =)
[14:59:23] meffie joins the room
[15:00:56] <wiesand> Hello
[15:01:02] <jgorse> Hello!
[15:02:12] <wiesand> I expect this meeting to be a brief one, especially w/o Ben
[15:02:30] <jgorse> fair enough
[15:02:41] <meffie> good day.
[15:02:47] <wiesand> Looks like we have a first stab at Linux 4.10 in https://gerrit.openafs.org/#/c/12506/
[15:02:51] <meffie> yes.
[15:03:40] <meffie> needs review
[15:03:42] <wiesand> I know it's not ready for general inclusion. But do we know whether that's the only issue to address for 4.10 (yet)?
[15:04:19] <meffie> it's the only one i know of.
[15:04:56] <meffie> mvita: do know of others issues for 4.10
[15:05:20] <mvita> oh, hey, sorry, I was … elsewhere....
[15:05:49] <mvita> yes, that's the only known issue for 4.10
[15:06:00] <wiesand> Sounds good, thanks! Looks like releasing 1.6.21 in time for 4.10 is still feasible then.
[15:06:18] <wiesand> We should aim for a pre1 in a fortnight.
[15:06:18] <mvita> please remember that I only check buildability - I do NOT test under the rcs
[15:06:43] <mvita> maybe jgorse could try it out?
[15:07:02] <jgorse> yes!
[15:07:05] <jgorse> will do
[15:07:22] <wiesand> That would be great
[15:07:25] <wiesand> Thanks!
[15:07:46] <mvita> maybe I can sit in on that so I can learn how to test under new linux releases as well
[15:09:16] <mvita> Stephan, I wanted to ask specifically if you can regression test the 4.10 fix with your getcwd setup
[15:09:39] <mvita> you can do that on any kernel release from 3.18 and up
[15:09:45] <mvita> if you have one
[15:10:14] <mvita> don't bother testing older than that, it probably won't even build as noted.
[15:10:31] <wiesand> Not right now, but I'll test the "shake harder" stack before merging, so adding that would be feasible.
[15:10:39] <mvita> in fact, even before my 4.10 fix, that code was not correct under anything older than 3.18
[15:10:52] <mvita> because it presumed that d_invalidate returned void
[15:11:19] <mvita> shake harder would be useful too
[15:11:35] <mvita> so this brings up a topic I wanted to discuss w/ the group
[15:11:46] <wiesand> go ahead?
[15:12:07] <mvita> just how far back (linux-wise) do we want to provide support in 1.6x and in 1.8x and in master?
[15:12:22] <mvita> it's been a few years now since we cut 2.4 loose
[15:12:31] <mvita> linux24, that is
[15:12:57] <mvita> is it time to think about linux26 and linux28 sunset?
[15:13:07] <jgorse> will 26 ever die?
[15:13:20] <wiesand> Linux 2.4 is in theory still supported in 1.6 I think.
[15:13:22] <mvita> i don't know…
[15:14:29] <wiesand> 2.6 (+ tons of backports) is the RHEL6 kernel. Four more years...
[15:14:32] <meffie> there are versions of 2.6 that pre-date git.
[15:15:22] <meffie> i believe 2.6.11 and earlier are only in cvs.
[15:15:28] <mvita> meffie: true, but I'm not sure I see what you are driving at
[15:15:49] <meffie> we should sunset support for those in 1.8.
[15:16:03] <mvita> ahh, yeah, okay
[15:16:21] <meffie> ben mentions this in one of the gerrits (i dont have the number off hand)
[15:18:46] <wiesand> We shouldn't drop support for current enterprise distros in the 1.6 series if it can be avoided. That's RHEL6/7 (5 for just a few more weeks), Ubuntu 14.04/16.04 (12.04 for just a few more weeks), and SLES accordingly.
[15:18:57] <meffie> agreed
[15:18:59] <mvita> okay, so for now I need to fix the existing call to d_invalidate to work correctly - even without the 4.10 update (I almost split that gerrit in 2, now I wish I had - but I was in a hurry to get back to other stuff)
[15:20:04] <wiesand> I think breaking older ones which are EOL (like RHEL3/4) is ok when justified.
[15:20:11] <mvita> okay, thank you stephan - admins know this stuff, devs do not.
[15:20:32] <wiesand> Like, if coping with 4.10 w/o breaking those would need resources we just don't have.
[15:20:57] <mvita> no, it should be relatively easy
[15:21:29] <mvita> it just seemed an opportune time to raise the issue
[15:22:57] <mvita> I wish I had been an AFS admin or Linux admin before working on this project
[15:23:34] <wiesand> RHEL6 is 2.6.32 + tons of backports. That's probably the oldest kernel we'd want to make an effort to support at this point. Especially in 1.8.
[15:23:48] <wiesand> Mark: Don't ;-)
[15:23:54] <meffie> it's not too late ;)
[15:24:36] <mvita> meffie: okay, as long as I still get to look at cores
[15:25:28] <wiesand> Oh you will. Though as an admin you'll want to disable them altogether for security reasons.
[15:27:01] <wiesand> Looking at some older changes in the 1.6 queue...
[15:27:18] <wiesand> I'm still curious whether 12328 is something we've been missing.
[15:27:27] <mvita> looking
[15:28:10] <wiesand> Do we still want 11151 (presumably with a backport of 12455)?
[15:28:31] <mvita> 12328 makes my brain hurt
[15:29:11] <wiesand> comforting
[15:29:40] <meffie> mvita: we should talk to yadav about it.
[15:29:45] <mvita> 11151 yes, still wanted
[15:30:01] <mvita> vos opacity is a continual source of pain
[15:30:53] <mvita> (no admin experience required to know that one)
[15:31:20] <wiesand> I'll try to pull up 12455 then.
[15:31:56] <mvita> thank you!
[15:32:44] <wiesand> Other candidates for 1.6.21 are inumcalc/ONEGROUP_ENV (I think to help Solaris)
[15:32:58] <mvita> yes, please
[15:33:28] <wiesand> I know some of this was merged - will check before the next meeting and see where we are.
[15:33:45] <wiesand> I do remember path conflicts with the change harder stack, so that should be merged first before pulling up.
[15:34:17] <meffie> i am doing the backport/cherrypicks for the ONEGROUP stuff now.
[15:34:24] <wiesand> Same fore the fs getcache miscounts.
[15:34:35] <meffie> just retesting.
[15:34:42] <wiesand> Mike: won't help much unless you place them on top of shake hadrer.
[15:34:51] <wiesand> or anything else that might clash
[15:35:23] <meffie> i can rebase them on other gerrits if you like.
[15:35:57] <wiesand> hold your horses and nothing will clash and have to be rebased ;-)
[15:36:20] <meffie> i rebase stuff all the time. no biggie :)
[15:37:52] <meffie> yes, the ONEGROUP stuff is for solaris. it fixes pag usage on solaris 11
[15:40:03] <wiesand> Last request I noted for the wish list is the arm64_linux26 topic. But I wouldn't dig into that until the others are done.
[15:40:23] <meffie> excellent.
[15:40:35] <wiesand> Anything else that should urgently make it into a 1.6 release?
[15:40:54] <meffie> that's all i know about.
[15:41:42] <mvita> nothing here
[15:43:39] <wiesand> Looks like we have a plan then: 1.6.21 in line with Linux 4.10, pre1 in ~2 weeks, everybody please review and test 12506, and let's see which of the wish list items we can squeeze in too.
[15:45:04] <wiesand> (and include "shake harder" if it tests ok at least for the known cases)
[15:45:18] <wiesand> Thoughts?
[15:45:30] <meffie> sounds good.
[15:45:41] <mvita> agreed
[15:46:26] <wiesand> Anything else to discuss today?
[15:46:53] <meffie> i have nothing
[15:47:14] <mvita> nothing here
[15:47:47] <wiesand> Great. I'll go and marvel at the blizzard then :)
[15:47:57] <mvita> oh?
[15:48:43] <wiesand> No worries. I'm safe and sound, at home.
[15:49:15] <mvita> you should be snow-camping - meffie envies you your snow
[15:50:04] <mvita> but he would feel better if he knew you weren't letting all that snow go to waste
[15:50:33] <meffie> just got new sleeping pads, from sweden, for the winter camping.
[15:50:56] <mvita> told you
[15:51:59] <wiesand> It's not going to last long enough here.
[15:53:03] <wiesand> Unfortunately the winter months are technically problematic for me to take time off.
[15:54:18] <wiesand> Thanks a lot everyone for participating today!
[15:54:22] <jgorse> =)
[15:54:25] <mvita> okay, I've gotta go - thank you for keeping things going Stephan!
[15:55:03] <wiesand> *YOU* are keeping things going.
[15:55:15] <wiesand> Off to the window pane now. Bye.
[15:55:19] wiesand leaves the room
[16:29:27] meffie leaves the room
[16:51:54] jgorse leaves the room
[18:58:31] meffie joins the room
[21:36:31] jgorse joins the room