- Wednesday, February 15, 2017

release-team@conference.openafs.org

Wednesday, February 15, 2017< ^ >

Room Configuration

Room Occupants

GMT+0
[00:11:50] mvita leaves the room
[02:44:47] jgorse joins the room
[04:10:00] mvita joins the room
[04:18:39] mvita leaves the room
[14:16:06] meffie joins the room
[14:25:47] mvita joins the room
[15:01:02] wiesand joins the room
[15:01:45] <meffie> good day
[15:01:45] kadukoafs@gmail.com/barnowlE3E87EB2 leaves the room
[15:01:46] <wiesand> Hello
[15:02:33] kadukoafs@gmail.com/barnowlD058797B joins the room
[15:02:47] <mvita> hi
[15:02:49] <kadukoafs@gmail.com/barnowlD058797B> greetings
[15:02:58] <meffie> hello
[15:03:45] <wiesand> so, Linux 4.10 is at rc8
[15:03:55] <wiesand> which means we're not late yet…
[15:03:59] <mvita> right
[15:04:02] <meffie> yes.
[15:04:27] <wiesand> but then, 12506 is not yet final
[15:05:03] <kadukoafs@gmail.com/barnowlD058797B> indeed
[15:05:14] <mvita> we are working on it
[15:05:25] <wiesand> And btw, I won't be available next week ;-(
[15:05:26] <mvita> thank you mike for your review
[15:05:44] <meffie> mvita: do you agree with my comments/questions  there?
[15:06:03] <meffie> i can do the autoconf foo if that makes helps.
[15:06:08] <kadukoafs@gmail.com/barnowlD058797B> I'm not sure that HAVE_HAVE_SUBMOUNTS is the proper autoconf macro,
though ;)
[15:06:11] <mvita> the D_INVALIDATE_IS_VOID will still be needed
[15:06:22] <mvita> I have to get my head back in this today
[15:06:22] <meffie> HAVE_HAVE.. is strange. :)
[15:06:36] <mvita> then I will let you know about the have_submounts conditional
[15:07:04] <mvita> unfortunately 1) I am buried and 2) I am sick
[15:07:08] <meffie> i see have_submounts() in v4.9, but it's gone in v4.10rcX
[15:07:19] <mvita> so I can't promise when I can have a new patchset
[15:07:30] <mvita> right
[15:07:32] <wiesand> get well soon
[15:07:44] <mvita> but that's not the real issue, as explained in the commit msg
[15:07:49] <kadukoafs@gmail.com/barnowlD058797B> Yes, get well soon
[15:07:58] <meffie> yes, get well, we can look at the patch.
[15:08:18] <meffie> luckly, 4.10 is not final yet.
[15:08:31] <mvita> we should not have been using have_submounts for a long time, long before it actually went away
[15:08:53] <meffie> sure, but that's water on the bridge.
[15:09:07] <meffie> under the bridge
[15:09:21] <mvita> so the remaining effort is to make sure this also works for pre 3.x
[15:09:29] <mvita> that's what I need to look at
[15:09:40] <meffie> well...
[15:09:57] <meffie> that can be done in a second patch.
[15:10:13] <mvita> hmm
[15:10:20] <meffie> i mean, we are already running the current code on 3.x.
[15:10:42] <meffie> changing that is a different change
[15:10:54] <wiesand> we're also running it on 2.6.x
[15:11:00] <mvita> oh?
[15:11:21] <mvita> I wasn't sure at all that 2.6.x would work correctly
[15:11:28] <wiesand> Er, I mean 1.6 still supports 2.6.x and it's used on those kernels.
[15:11:41] <meffie> yes, it is.
[15:12:12] <mvita> <wishes again that this meeting had audio>
[15:12:16] <wiesand> So whether it's a second patch or included in 12605, we need the change before 1.6.20.2 can be released.
[15:12:42] <mvita> okay
[15:12:52] <meffie> i am not so sure that is true.
[15:14:47] <meffie> seems we should have 12605 so we can use 4.10 and beyond.
[15:15:29] <meffie> and it should build the same code we always had if we still have_submounts().
[15:15:29] <wiesand> but can we still use 2.6.x ? (let a lone 2.4.x)
[15:15:59] <mvita> I.  Don't.  Know.  Yet.
[15:16:10] <kadukoafs@gmail.com/barnowlD058797B> I suppose you could try a runtime test :)
[15:16:14] <mvita> I will investigate and let you know what I find.
[15:16:28] <meffie> i would think it conditionals would allow it to build the same code as before, that's just mechanical.
[15:16:46] <meffie> that's all i mean.
[15:17:36] <meffie> and if we want to change how it works in the old versions, that could be a different patch.
[15:17:59] <wiesand> ah
[15:18:05] <mvita> brb
[15:18:21] <wiesand> ?
[15:18:31] <meffie> brb == be right back
[15:18:37] <wiesand> ah…
[15:18:41] <kadukoafs@gmail.com/barnowlD058797B> Sounds like mark got a phone call or something
[15:19:21] <wiesand> re "try a runtime test": unfortunately I'm buried too…
[15:20:13] <meffie> can we just use a compile time test to see if have_submounts() is still available?
[15:21:53] <meffie> well, some more work needs to be done.
[15:22:31] <wiesand> finally a statement even I can understand
[15:22:34] <meffie> i'll try to get more people to review it.
[15:24:00] <wiesand> thanks
[15:24:22] <wiesand> what else to discuss today? 1.8?
[15:24:32] <meffie> i have a buildbot update
[15:24:59] <wiesand> ah :)
[15:25:06] <mvita> I'm back
[15:25:43] <kadukoafs@gmail.com/barnowlD058797B> Joe will be happy that I started looking at 11794
[15:25:57] <meffie> per ben, i've promoted the win7 builders from nightly to gerrit triggered. all 4 of them are now gerrit triggered.
[15:26:14] <meffie> i've started adding more build slaves too.
[15:26:41] <kadukoafs@gmail.com/barnowlD058797B> But I had some questions about 11794 that I could not resolve myself.
[15:26:47] <meffie> i've added a centos73 builder (nightly) and an ubuntu16.10 (nightly) so far.
[15:27:18] <mvita> when did we get windows builders again?
[15:27:46] <meffie> some weeks ago. hosted at MIT
[15:27:51] <kadukoafs@gmail.com/barnowlD058797B> December
[15:28:01] <mvita> somehow I missed that - oh, vacation.
[15:29:09] <meffie> i also added a new centos 7 builder that does a nightly "make check".  it doesnt pass yet. :)
[15:29:36] <kadukoafs@gmail.com/barnowlD058797B> > make check ... doesn't pass
ugh.  Thanks for getting that started, at least.
[15:29:42] <kadukoafs@gmail.com/barnowlD058797B> We also had a couple of non-spam things hit RT.
[15:30:10] <meffie> ok.
[15:30:27] <kadukoafs@gmail.com/barnowlD058797B> Stephan, could you understand what was going on with the guy that had
rpms and was getting weird issues at startup?  I was confused about
both where his rpms were coming from and how using yum to install
would have helped.
[15:30:51] <kadukoafs@gmail.com/barnowlD058797B> In particular, I only saw a src.rpm on the page he linked, when I
looked.
[15:31:13] <wiesand> let me check...
[15:31:34] <meffie> oh, yes, that was odd.
[15:32:10] <meffie> (charles yan was posting the the cmu afs list.)
[15:33:06] <kadukoafs@gmail.com/barnowlD058797B> Oh, I didn't mean you needed to go read it all right now; I just
wondered if you saw it as it came in.
[15:34:48] <wiesand> I saw it, couldn't make much sense of it… I'm really buried…
[15:34:51] <meffie> i missed that he also submitted a ticket. i just was trying to help him on the mail list.
[15:35:16] <wiesand> Could be because I'm not quite up to date uploading the latest kernel module packages
[15:35:51] <kadukoafs@gmail.com/barnowlD058797B> Well, I ... guess it's good that it's not just me who was having
trouble making sense of it.
[15:36:07] <kadukoafs@gmail.com/barnowlD058797B> meffie: is that list archived publicly?
[15:36:09] <meffie> not just you.
[15:36:27] <meffie> hmm, dont know.
[15:36:29] <wiesand> It's a bit tedious. I can either rsync everything built for all the prehistoric kernels, or pcik the somewhat current ones and do the metadata manually.
[15:37:10] <wiesand> Regarding http://www.openafs.org/release/openafs-1.6.20.html
[15:37:45] <wiesand> We release w/o binaries. So, no binaries visible on the release page. This the "All files" link.
[15:38:12] <wiesand> Updating that page is tedious too. Especially if you'd want ot do it every time a new kernel module is uploaded.
[15:38:26] <kadukoafs@gmail.com/barnowlD058797B> Ah, that's the part I was missing/forgetting.
[15:39:25] <meffie> btw, that list == info-afs@grand.central.org
[15:40:31] <meffie> https://lists.central.org/mailman/listinfo/info-afs
[15:40:34] <wiesand> -642.el6 is suffiicently old though… that module is available on the server
[15:40:36] <kadukoafs@gmail.com/barnowlD058797B> thanks
[15:40:39] <meffie> yes, it is archived.
[15:41:53] <kadukoafs@gmail.com/barnowlD058797B> Anyway, the other RT entry that I remember is someone trying out 1.8
with transarc paths and having lots of trouble.
[15:42:03] <wiesand> but then yum picking the right kernel module when resolving dependencies is… well, it may happen… but it rarely does
[15:42:08] <kadukoafs@gmail.com/barnowlD058797B> But, I also had a hard time following this person's messags (though
for different reasons).
[15:42:15] <mvita> ted creedon
[15:42:59] <mvita> agreed, both tickets were hard to follow
[15:43:28] <kadukoafs@gmail.com/barnowlD058797B> But, it probably is worth having someone test 1.8 with transarc paths
to see if we can get some clearer data.
[15:43:33] <kadukoafs@gmail.com/barnowlD058797B> Any volunteers?
[15:44:03] <mvita> <avoids eye contact>
[15:44:13] <wiesand> <whistles>
[15:44:21] <meffie> well, test 1.8 with transarc paths all the time
[15:44:24] <meffie> i test
[15:44:38] <kadukoafs@gmail.com/barnowlD058797B> :)
[15:45:10] <meffie> i know it's deprecated, but old habits die hard. ;)
[15:46:20] <wiesand> A while ago I was asked "are non-transarc paths very different?". I won't tell by whom.
[15:46:50] <meffie> i took ted's ticket, 134130: 1.8.pre1 akeyconvert fails
[15:47:02] <kadukoafs@gmail.com/barnowlD058797B> thanks
[15:47:44] <kadukoafs@gmail.com/barnowlD058797B> I feel like there's also some weirdness with 1.8 clients going on that
has only partially been reported/characterized, like sometimes it just
fails to access certain paths or something.
[15:48:52] <meffie> hmm.
[15:49:45] <kadukoafs@gmail.com/barnowlD058797B> I have probably seen something like that but was in too much of a
hurry to try and characterize it better, and there was at least one
user report.
[15:51:55] <meffie> ok. well, hopefully we can get more clues.
[15:52:46] <wiesand> NB running the client on EL5/6/7 with SELinux enabled and enforcing generally works fine.
[15:53:21] <kadukoafs@gmail.com/barnowlD058797B> wiesand: the 1.6 client, you mean?
[15:53:50] <wiesand> The SL packaging has an explicit "chcon -t mnt_t /afs" in %post though, so at least at some point in the past that was needed.
[15:54:04] <wiesand> Yes: 1.6. I'm still looking at that ticket.
[15:54:19] <kadukoafs@gmail.com/barnowlD058797B> I was pretty sure, just wanted to check.
[15:54:59] <wiesand> The server is a different story. Since EL6 the policy knows about AFS, but it's out of date.
[15:55:19] <wiesand> Doesn't know about fssync sockets for example.
[15:56:28] <wiesand> Sorry for the 1.6 distraction
[15:57:09] <kadukoafs@gmail.com/barnowlD058797B> no worries
[15:58:58] <wiesand> At least we now have confirmation that there's at least one user of our binary packages left.
[15:59:09] <meffie> yes
[16:00:30] <wiesand> I'll upload the latest kernel modules later today then…
[16:00:43] <meffie> thanks. motion to adjourn?
[16:01:05] <wiesand> sustained, unless anyone...?
[16:01:49] <wiesand> It seems not. Let's adjourn.
[16:01:54] <wiesand> Thanks a lot everyone!
[16:02:07] <meffie> thank you! stay safe.
[16:02:43] wiesand leaves the room
[16:03:15] meffie leaves the room
[18:55:55] mvita leaves the room