- Wednesday, February 1, 2017

release-team@conference.openafs.org

Wednesday, February 1, 2017< ^ >

Room Configuration

Room Occupants

GMT+0
[14:51:47] jgorse leaves the room
[14:58:09] <mvita> <sneaks in early>
[14:58:58] Stephan Wiesand joins the room
[14:59:02] <kadukoafs@gmail.com/barnowlE3E87EB2> How sneaeky.
[14:59:23] <Stephan Wiesand> hey, I'm almost aminute eraly tooo
[14:59:43] <kadukoafs@gmail.com/barnowlE3E87EB2> You can be sneaky, too.
[15:01:26] jgorse joins the room
[15:02:51] <Stephan Wiesand> Ok, bad news first: My time budget for openafs is practically zero currently.
[15:03:38] <mvita> but you still have lots of vacation time, right ?  ;-)
[15:03:41] <Stephan Wiesand> (and Telekom stalls my network traffic for O(30s) frequently :-(((
[15:04:21] <Stephan Wiesand> not really
[15:04:21] <Stephan Wiesand> Mark: in theory
[15:04:35] <kadukoafs@gmail.com/barnowlE3E87EB2> I am also losing some time since the TLS certs for the MIT machine
expire in a month and I need to sort out replacements.
[15:05:19] meffie joins the room
[15:06:33] <meffie> letsencrypt?
[15:06:42] <kadukoafs@gmail.com/barnowlE3E87EB2> Stephan, are there things that the rest of us could do that would help
you?
[15:07:01] <kadukoafs@gmail.com/barnowlE3E87EB2> Yeah, it may end up being letsencrypt.
[15:07:19] <Stephan Wiesand> So, due to the time shortage and the fact that much of what was discussed isn't ready, I propose a change of plans: 1.6.20.2 about int time for Linux 4.10. Maybe some other platform specific stuff liek macos prefpane. Postpone anything else to a later 1.6.21.
[15:08:07] <kadukoafs@gmail.com/barnowlE3E87EB2> I don't think I am quite bold enough to suggest skipping 1.6.21
entirely in favor of 1.8.0 ... yet ;)
[15:09:19] <mvita> stephan, I think that's fine
[15:09:43] <meffie> sounds good.
[15:09:55] <kadukoafs@gmail.com/barnowlE3E87EB2> But yes, that sounds fine.
[15:11:05] Stephan Wiesand leaves the room
[15:11:14] Stephan Wiesand joins the room
[15:11:33] <Stephan Wiesand> and my connection is terrible today
[15:12:02] <Stephan Wiesand> Mark, Mike: thanks for your support
[15:12:13] <meffie> btw, buildbot update. i contacted the admins for the bots that were down, and i made my first actual change to the buildbot master config.
[15:12:14] <mvita> we appreciate all you do
[15:12:23] <Stephan Wiesand> Ben: no, not yet. But I feel it can wait another Linux cycle
[15:13:02] <kadukoafs@gmail.com/barnowlE3E87EB2> Exciting news, Mike!
[15:14:09] Stephan Wiesand leaves the room
[15:14:28] Stephan Wiesand joins the room
[15:14:31] <meffie> yeah, so i will next add some new nightlies and then look at making a gerrit vs nightly waterfall.
[15:15:09] <Stephan Wiesand> so, the one change we really need is the Linux 4.10 one, not yet merged on master
[15:15:15] <kadukoafs@gmail.com/barnowlE3E87EB2> W.r.t linux 4.10, it sounds like 12506 is believed to be the right
thing for 3.18+, but may break the older linux versions?
[15:16:03] <mvita> oh?   I don't know.
[15:16:34] <mvita> does it need to be reworked?
[15:17:18] <kadukoafs@gmail.com/barnowlE3E87EB2> You left a comment on it early on "This will NOT work correctly (and
may not even build) on older releases of Linux (pre 3.18) where
d_invalidate still could return -EBUSY".
[15:17:43] <mvita> oh, I had forgotten about that
[15:17:47] <Stephan Wiesand> it builds on EL5, EL6 and EL7
[15:18:20] <Stephan Wiesand> but I haven't found time to deploy those test builds :-(
[15:18:50] <mvita> stephan, what was that test you would run to provoke the getcwd issue?
[15:19:16] <mvita> I think that might be a good test for this (provoking -EBUSY, that is)
[15:19:59] <Stephan Wiesand> git clone openafs; git show IIRC
[15:20:46] <jgorse> Mark:
[15:20:46] <jgorse> https://hastebin.com/suzipaguta.bash
[15:20:54] <mvita> Ben, thank you for the ping, I will try to come up with another patch to support all Linux releases
[15:21:12] <kadukoafs@gmail.com/barnowlE3E87EB2> Thanks
[15:21:59] Stephan Wiesand leaves the room
[15:22:12] Stephan Wiesand joins the room
[15:23:00] <kadukoafs@gmail.com/barnowlE3E87EB2> While we're scrounging for volunters, if someone wanted to look at the
-devel thread and submit patches for the extra semicolon and a
configure option to disable swig/perluafs, that would probably free up
more of my time for things like reviewing 11794...
[15:23:39] <meffie> i can blame andrew for the swig stuff and see if he can help.
[15:23:55] <mvita> good plan
[15:24:02] <mvita> cmarko could be plan b
[15:24:12] <kadukoafs@gmail.com/barnowlE3E87EB2> What's weird is that the swig binary segfaults; it doesn't give us any
information about what it doesn't like.
[15:24:27] <kadukoafs@gmail.com/barnowlE3E87EB2> But there's got to be some other variable, since I couldn't reproduce
it on my debian VM.
[15:25:00] <kadukoafs@gmail.com/barnowlE3E87EB2> And we probably should have a configure knob to forcibly disable it,
independently of fixing the swig part.
[15:25:03] <mvita> ug
[15:25:17] <meffie> if i recall, the swig but hits on SuSE?
[15:25:22] <meffie> bug
[15:25:25] <mvita> (ug for the unpredictability)
[15:25:36] <kadukoafs@gmail.com/barnowlE3E87EB2> The SuSE builder has been hitting it, yeah.  And this report was on
solaris.
[15:27:48] <meffie> ok, thanks.
[15:28:18] <meffie> i can add the configure knob, regardless. that is a good idea.
[15:30:28] <kadukoafs@gmail.com/barnowlE3E87EB2> Great; thanks
[15:31:36] <kadukoafs@gmail.com/barnowlE3E87EB2> <pumpkin>
[15:31:52] Stephan Wiesand leaves the room
[15:33:10] Stephan Wiesand joins the room
[15:33:31] <Stephan Wiesand> I think I'll call it a day. Connection is just unusable
[15:34:04] <mvita> ok,
[15:34:38] <Stephan Wiesand> (I'm using what's supposed to be our premium provider - it used to be ok, but recently...)
[15:35:20] <Stephan Wiesand> Famous last words: we need the Linux 4.10 thingy, I won't be able to help much.
[15:37:06] <Stephan Wiesand> If we can't muster the resources to cope, we should probably denounce support for future Linux clients :-(
[15:38:47] <mvita> I think we're keeping up okay.  Sorry I forgot about the pre-3.18 consideration for this.
[15:39:07] <mvita> I had convinced myself it was done
[15:42:00] <Stephan Wiesand> (sorry, that probably was my general frustration speaking)
[15:42:04] <meffie> hopefully jgorse will have more time soon too.
[15:42:28] <jgorse> what is the issue?
[15:42:47] <meffie> i mean after the perf stuff.
[15:43:02] <jgorse> ah… yes… slow churn, but better ice cream
[15:43:18] <Stephan Wiesand> ?
[15:43:31] <jgorse> the going is slow, but it is good
[15:43:37] <jgorse> it needs to be done
[15:44:25] <Stephan Wiesand> thanks
[15:45:37] <kadukoafs@gmail.com/barnowlE3E87EB2> Everyone saw the IRC traffic about a potential kafs hackathon in
Cambridge, MA, right?
[15:46:19] <Stephan Wiesand> No...
[15:46:21] <mvita> yes
[15:46:35] <Stephan Wiesand> Is that good news or bad?
[15:46:53] <jgorse> Good
[15:47:10] <mvita> sorry, gotta go to another appt
[15:47:37] <jgorse> ttyl
[15:47:40] <Stephan Wiesand> (according to Jeffrey kafs clients will kill openafs servers like flies; there are open tickets on these issues)
[15:48:05] <Stephan Wiesand> bye marl
[15:48:47] <jgorse> Kill openafs servers?
[15:48:55] <meffie> yeah
[15:49:19] <Stephan Wiesand> yeah :-(
[15:49:25] <meffie> we have some bugs to fix.
[15:50:17] <jgorse> ah… fair enough
[15:50:32] <Stephan Wiesand> thus, mixed feelings :-/
[15:51:32] <jgorse> it is good to fix bugs =)
[15:52:49] <Stephan Wiesand> s/is/would be/
[15:53:27] <jgorse> they must be discovered/cause pain before they can be fixed. it is normal part of the process.
[15:56:06] <Stephan Wiesand> Well, they are discovered. It's hard to predict when they'll strike.
[15:58:04] <Stephan Wiesand> Worst case, another AFS implementation updates their RX to behave like the kafs one.
[16:00:34] <Stephan Wiesand> That would turn openafs into something you can only use on trusted networks and with trusted clients. Over night.
[16:02:33] <Stephan Wiesand> But I really ought to stop spreading pessimism here. Sorry for my current mood.
[16:02:51] <Stephan Wiesand> Let's adjourn, cu next week.
[16:02:54] Stephan Wiesand leaves the room
[16:04:31] meffie leaves the room
[21:01:33] mvita leaves the room
[21:23:28] mvita joins the room
[21:45:23] mvita leaves the room