Home
release-team@conference.openafs.org
Friday, August 9, 2019< ^ >
Room Configuration
Room Occupants

GMT+0
[00:06:13] yadayada leaves the room
[01:33:36] yadayada joins the room
[02:14:38] yadayada joins the room
[02:30:28] yadayada leaves the room
[02:46:55] yadayada leaves the room
[02:55:27] yadayada joins the room
[04:55:32] yadayada leaves the room
[12:02:50] mbarbosa joins the room
[12:56:54] meffie joins the room
[14:53:35] yadayada joins the room
[14:57:39] <meffie> good day.
[14:58:19] <kaduk@jabber.openafs.org/barnowl> greetings
[14:58:34] <yadayada> hello everyone
[15:00:57] wiesand joins the room
[15:01:10] <wiesand> hello
[15:01:15] mvita joins the room
[15:01:20] <mvita> hello
[15:02:19] <wiesand> re 1.8.4pre1: it's on its way… slowly, but it is
[15:02:53] <wiesand> just deployed a last(?) test build with everything we agreed on to include
[15:03:21] <kaduk@jabber.openafs.org/barnowl> Yay!
[15:03:57] <wiesand> that test build also includes the Linux 5.3 changes from master (13752 13753)
[15:04:01] <meffie> yay!!
[15:04:32] <wiesand> of course it's running on EL7, so those shouldn't actually matter - but at least they don't break anything
[15:05:00] <kaduk@jabber.openafs.org/barnowl> Not breaking anything is important :)
[15:05:15] <mvita> NOW you tell me.
[15:05:30] <wiesand> frankly, i actually consider it more important than amking bleeding edge kernels work
[15:06:06] <wiesand> now, given that Linux 5.3 is is fairly close, should we have those two changes in 1.8.4pre1?
[15:06:29] <wiesand> or issue that now, and either have a pre2 or 1.8.4.1 soon?
[15:07:35] <wiesand> and btw are those sufficient to make our client work on 5.3, or is there more to dO/
[15:07:41] <mvita> there may be more
[15:07:52] <mvita> Cheyenne is investigating a new issue now
[15:08:12] <mvita> it was just discovered so I have no further details
[15:08:42] <mvita> nothing to do with the first set of 5.3 work
[15:08:51] <mvita> it's in the keyring API
[15:08:55] <kaduk@jabber.openafs.org/barnowl> Sounds like 1.8.4.1 might be the more prudent path to aim for
[15:09:18] <wiesand> ok, thanks… let's not block on 5.3 then, agreed
[15:10:10] <yadayada> nothing much from my side this week, our dput fix 13748 is working fine in of ur customer env, where deadlock was reported. Sent some aklog memory leaks to Ben and Andrew.  I need to test 13705 (native mount support) in 1.6 code base I saw softlock when root volume was mounted at multiple places.
[15:10:19] <wiesand> in other news, I smoke tested 1.6.23 with the RHEL7.7 kernel and it seems there's nothing to do :)
[15:10:53] <kaduk@jabber.openafs.org/barnowl> Apparently I never packaged 1.8.3 for Debian and now Ubuntu has a new
enough kernel to need it :(
[15:12:22] <wiesand> Chances are Anders will add it to the ppa then, no?
[15:13:00] <kaduk@jabber.openafs.org/barnowl> Anders tends to take the packaging from Debian and put it in the PPA
[15:13:12] <wiesand> The "official" Ubuntu 18.04 packages are still 1.8.0pre… they're unusable anyway
[15:14:18] <kaduk@jabber.openafs.org/barnowl> Yeah, it's kind of unfortunate.
[15:14:45] <kaduk@jabber.openafs.org/barnowl> but I don't have the cycles to do anything about it, and the process
is somewhat annoying, too
[15:14:47] <wiesand> If we get out 1.8.4pre1 really soon now, would that be a candidate for debian -> ubuntu ppa ?
[15:15:22] <kaduk@jabber.openafs.org/barnowl> Yeah, that would probably be fine; we just released buster so I'm not
worried about having a beta in unstable or testing for a bit
[15:15:42] <kaduk@jabber.openafs.org/barnowl> But it's the same amount of effort to package 1.8.3 and 1.8.4pre1 (and
in either case shouldn't be much effort)
[15:18:02] <wiesand> Up to you of course. Some real world exposure of 1.8.4pre1 would be rather nice though
[15:18:16] <kaduk@jabber.openafs.org/barnowl> True
[15:18:58] <wiesand> while I finalize NEWS, want to move on to master/1.9?
[15:19:36] <kaduk@jabber.openafs.org/barnowl> I didn't get much done, but it looks like other people did, at least
[15:20:10] <kaduk@jabber.openafs.org/barnowl> Yadav mentioned the memory leak he's working on, and the linux 5.3
stuff that you saw already
[15:20:27] <kaduk@jabber.openafs.org/barnowl> I'm surprised that you haven't said anything about the "rx_dead" topic
yet, though ;)
[15:21:59] <mvita> oh, my apologies about the 5.3 issue, it's already addressed in https://gerrit.openafs.org/#/c/13752
[15:22:15] <mvita> I mistakenly thought all the 5.3 fixes were already merged on master
[15:22:20] <kaduk@jabber.openafs.org/barnowl> (On first look it does not seem like it can have any fateful
interactions with idledead)
[15:22:50] <mvita> correct, orthogonal with idledead
[15:23:26] <mvita> in fact, the hang can only happen when idle timeout is off, e.g. volservers
[15:23:31] <kaduk@jabber.openafs.org/barnowl> mvita: understandable to think they were merged, since I +2'd that
one.
[15:23:45] <kaduk@jabber.openafs.org/barnowl> But I maybe want to quibble about a nit on the other 5.3 one
[15:24:36] <wiesand> rx_dead did sound my alarm bell ;-)
[15:26:14] <wiesand> glad to hear it's indeed not about idledead
[15:26:57] <mvita> I specifically avoided fixing this by turning on idledead for volservers - you are always in my thoughts, Stephan!
[15:30:02] <wiesand> Thanks :)
[15:30:38] <kaduk@jabber.openafs.org/barnowl> Andrew looked at the macOS notarization stuff and asked some good
questions, so I'll look forward to seeing them answered
[15:30:45] <wiesand> clientsideideledeadistherootofallevil
[15:30:47] <wiesand> clientsideideledeadistherootofallevil
[15:30:48] <wiesand> clientsideideledeadistherootofallevil
[15:30:50] <wiesand> ...
[15:31:02] <meffie> heh
[15:33:57] <wiesand> (hopefully) final NEWS pushed to 13643
[15:34:31] <meffie> woot!
[15:34:50] <meffie> i'll update the 1.9.0 news gerrit.
[15:36:16] <kaduk@jabber.openafs.org/barnowl> Thanks
[15:37:40] <wiesand> if all agree, i could merge that and 13596 now, and then Ben could tag it - making it eligible for debian i believe
[15:39:29] <kaduk@jabber.openafs.org/barnowl> Seems reasonable (looking at NEWS now)
[15:39:41] <mvita> looking at NEWS now as well
[15:41:55] <meffie> i just looked at the last diff since my last +1. lgtm
[15:42:58] <kaduk@jabber.openafs.org/barnowl> I have a grammar nit but it can be fixed for the final 1.8.4 and
doesn't need to happen now, also +1
[15:46:57] <mvita> +1
[15:48:21] <wiesand> I kind of stumbled over that sentence too - will keep your proposed change in mind for 1.8.4 final.
[15:48:47] <wiesand> About to merge the final 2 changes for 1.8.4pre1 now
[15:49:34] <wiesand> unless  yadav would like to have some time for examining them, that is
[15:50:41] <yadayada> sure I will go through it and let you know my comments if any
[15:52:38] <wiesand> thanks, but to be sure: any objections to merging them now and thus get 1.8.4pre1 tagged during this meeting?
[15:53:07] <mvita> re:  https://gerrit.openafs.org/#/c/13643, I'm reminded that I now have two gerrit accounts:  "Mark Vitale" (Ubuntu launchpad openid) and "Mark Alternate Vitale"  (yahoo openid).  Have any of you succeeded in having multiple authenticators under one gerrit ID?  If so, could you share some tips?
[15:53:37] <kaduk@jabber.openafs.org/barnowl> You have to add a new authenticator to an existing account without
trying to log in using it first.
[15:53:46] <kaduk@jabber.openafs.org/barnowl> (Or, get some backend db surgery done.)
[15:53:58] <wiesand> "pester Ben to fix the DB for you" ;-)
[15:54:31] <mvita> I don't like to pester Ben, he's so busy.
[15:54:57] <meffie> wiesand: no objections
[15:55:01] <mvita> it's not that important
[15:56:02] <kaduk@jabber.openafs.org/barnowl> You could skim through
https://gerrit.openafs.org/Documentation/cmd-index.html and similar to
see if there's native "merge accounts" functionality
[15:56:15] <mvita> okay, thanks
[15:56:21] <kaduk@jabber.openafs.org/barnowl> I vaguely remember seeing something like that, but maybe it was for a
newer version than we're running :(
[15:59:22] <kaduk@jabber.openafs.org/barnowl> It sounds like we might be winding down, other than Stephan's question
to Yadav.
[16:00:03] <yadayada> looks good to me ..
[16:00:23] <meffie> hooray!
[16:00:24] <kaduk@jabber.openafs.org/barnowl> I'll note that we have a few changes with code review +2 but no
buildbot feedback; maybe someone with some spare time could try to
trigger some builds for them
[16:01:06] <meffie> i am going to trigger builds for all of them
[16:01:14] <kaduk@jabber.openafs.org/barnowl> Cool, thank you
[16:01:33] <meffie> i can change the sort order to do the higher numbers first.
[16:01:43] <kaduk@jabber.openafs.org/barnowl> (I also don't know what's up with the two four-digit changes that it's
been failing to build recently.)
[16:02:03] <kaduk@jabber.openafs.org/barnowl> Make sure you do an integer comparison and not a string comparison :)
[16:02:06] <wiesand> ok, changes arrived in git - I think we'ready to tag 6ab27a5cf3292a20ab807b7bc4dca43f305f14de as openafs-stable-1_8_4pre1
[16:02:07] <meffie> that's on me. i have a command line tool that was stuck retrying.
[16:02:31] <meffie> yes, sort -n is my friend.
[16:02:53] <kaduk@jabber.openafs.org/barnowl> (I spent some time a couple weeks back convincing python to do the
right thing)
[16:03:13] <meffie> ++ force-gerrit-build.py 13719
[16:03:56] <meffie> it uses python requests to send the POST to buildbot.openafs.org. sorry about the python stephan.
[16:04:57] <kaduk@jabber.openafs.org/barnowl> tag pushed
[16:05:05] <meffie> yay!
[16:05:13] <wiesand> I was surprised to see it attempt to rebuild the antique 11678
[16:05:59] <meffie> the oldest ones are in the 1000 range.
[16:06:08] <kaduk@jabber.openafs.org/barnowl> *nods*
[16:08:10] <wiesand> I'm not religious about python, I just think it should be banned from anything serious...
[16:08:23] <meffie> which tag was pushed? i'm not seeing it yet.
[16:08:49] <meffie> oh, there it is.
[16:08:56] <kaduk@jabber.openafs.org/barnowl> * [new tag]               openafs-stable-1_8_4pre1 -> openafs-stable-1_8_4pre1
[16:08:57] <wiesand> be patient, it takes a minute or two
[16:08:58] <meffie> From git://git.openafs.org/openafs
* [new tag]             openafs-stable-1_8_4pre1 -> openafs-stable-1_8_4pre1
[16:10:45] <wiesand> NB I learned that EL8 won't have a /usr/bin/python, just python2 and/or python3 :-)
[16:11:11] <wiesand> Speaks volumes, doesn't it...
[16:11:11] <mvita> re:  https://gerrit.openafs.org/#/c/13684/  I discovered that this breaks the build on any old gcc that doesn't have -frame-larger-than
[16:11:34] <mvita> specifically in our internal build farm:  SLES 11.3
[16:11:40] <mvita> how much do we care?
[16:11:54] <mvita> that release is old, but still on LTS
[16:12:54] <kaduk@jabber.openafs.org/barnowl> We probably should care.  Sigh.
[16:12:58] <wiesand> does this affect the default gcc on EL7/6/5?
[16:13:09] <mvita> IDK
[16:13:16] <kaduk@jabber.openafs.org/barnowl> I guess we can do a configure check for that warning and
conditionalize the makefile.in
[16:13:59] <wiesand> and BTW, I think 13753 will break us on EL4 and earlier
[16:14:21] <kaduk@jabber.openafs.org/barnowl> What kernel does EL4 have?
[16:14:52] <mvita> it should work wiesand, send_sig is quite old
[16:15:06] <mvita> it's in the very first git commit for Linux
[16:15:23] <kaduk@jabber.openafs.org/barnowl> Unfortunately we still claim to support some pre-git kernels
[16:15:23] <wiesand> EL4 has 2.6.9 - I haven';t checked whether send_sig was backported
[16:15:34] <mvita> ah
[16:15:54] <meffie> rhel 4 had 2.6
[16:16:04] <mvita> well, it certainly could be older that that, we just have no way of telling
[16:16:17] <mvita> without actually checking headers or trying a build
[16:17:13] <mvita> it shouldn't matter - if autoconf fails for presence of send_sig, we fall back to the old behavior
[16:17:26] <mvita> so no more broken than it (might) be already
[16:17:52] <kaduk@jabber.openafs.org/barnowl> I thought the version in gerrit used send_sig unconditionally (no
configure test)
[16:18:10] <wiesand> there is no autoconf test in 13573, it switches to send_sig unconditionally
[16:18:37] <mvita> oh, you are correct, I'm remembering wong
[16:18:41] <mvita> wrong
[16:19:20] <wiesand> I don't actually object to going this route, just wanted everyone to be aware
[16:19:26] <kaduk@jabber.openafs.org/barnowl> But,
http://fxr.watson.org/fxr/source/kernel/signal.c?v=linux-2.4.22#L723
suggests that EL4 will be fine
[16:19:42] <mvita> when we discussed this issue internally we concluded that RH5 was the lowest we needed to support - is that incorrect?
[16:20:23] <kaduk@jabber.openafs.org/barnowl> (BTW, did lxr.linux.no stop getting maintained?  Their 2.5-2.6.11 link
failed for me.)
[16:20:40] <mvita> oh, I forgot about lxr
[16:21:40] <meffie> http://lxr.linux.no is up for me.
[16:21:49] <kaduk@jabber.openafs.org/barnowl> For me, too.
[16:22:07] <kaduk@jabber.openafs.org/barnowl> But the link to http://lxr.linux.no/linux-bk gives an internal server
error
[16:22:13] <meffie> oh, but yes, server errors
[16:22:46] <wiesand> I don't object to dumping support for obsolete linux kernels, it should be a conscious decision though
[16:23:19] <kaduk@jabber.openafs.org/barnowl> My recollection is that the current claim is to support 2.6 and newer,
but that we actively dropped 2.4.x and older
[16:23:46] <mvita> yes, 2.4 is dropped
[16:23:51] <wiesand> In fact this could pave the way for dumping ~75% of those configure checks, speeding up kernel module builds significantly
[16:24:08] <kaduk@jabber.openafs.org/barnowl> "configure is blazing fast on freebsd" :-P
[16:24:09] <mvita> so we are just talking about the nether world of 2.6.x pre-git
[16:24:49] <mvita> Cheyenne may be looking at pruning old Linux autoconf tests soon
[16:25:19] <wiesand> chances are that sites still running such old kernels aren't going to update the openafs client ever again either
[16:25:37] <wiesand> mark: that would be great!
[16:26:57] <wiesand> and shouldn't that be "blazing_ly_ fast" ?
[16:27:40] <mvita> It may even go to "11"
[16:30:07] <meffie> motion to adjourn?
[16:30:27] <meffie> (too hungry to continue to note take :)
[16:30:34] <wiesand> motion sustained
[16:30:45] <wiesand> thanks a lot everyone!
[16:30:45] <mvita> d'accord
[16:30:56] <yadayada> thanks
[16:31:02] <meffie> thanks all.
[16:31:55] <wiesand> stay tuned for 1.6.4pre1 uploads/web change/announcement draft
[16:32:07] <kaduk@jabber.openafs.org/barnowl> Thanks everyone!
[16:38:54] meffie leaves the room
[16:39:33] wiesand leaves the room
[16:59:55] yadayada joins the room
[17:10:34] yadayada leaves the room
[17:26:54] yadayada leaves the room
[17:42:25] yadayada joins the room
[18:02:13] yadayada joins the room
[18:14:28] mbarbosa leaves the room
[18:14:28] yadayada leaves the room
[18:17:34] yadayada leaves the room
[18:42:47] yadayada joins the room
[18:43:39] mbarbosa joins the room
[19:20:24] yadayada joins the room
[19:36:18] yadayada leaves the room
[19:52:17] yadayada leaves the room
[20:04:24] yadayada joins the room
[20:14:40] mbarbosa leaves the room
[20:22:39] yadayada joins the room
[20:30:32] yadayada leaves the room
[20:52:28] yadayada leaves the room
[21:27:48] mbarbosa joins the room
[22:48:56] mbarbosa leaves the room