- Friday, August 3, 2018

release-team@conference.openafs.org

Friday, August 3, 2018< ^ >

Room Configuration

Room Occupants

GMT+0
[00:16:37] mvita leaves the room
[01:14:05] mvita joins the room
[02:04:21] mvita leaves the room
[12:48:08] <kaduk@jabber.openafs.org/barnowl> (Hmm, did my client silently disconnect me?)
[12:52:12] mvita joins the room
[12:58:28] wiesand joins the room
[12:58:38] meffie joins the room
[12:59:39] <meffie> good morning.
[12:59:49] <kaduk@jabber.openafs.org/barnowl> greetings!
[12:59:56] <wiesand> good afternoon
[13:00:20] <wiesand> for a change, I have 1.6.x items
[13:00:43] <meffie> good afternoon
[13:01:05] <mvita> hello
[13:01:49] <kaduk@jabber.openafs.org/barnowl> Tell us about 1.6!
[13:02:03] <wiesand> I just built a test precursor of 1.6.23pre1. It has 12818 13029 13052 13053 13234 13261 13062 13083 13223 12895 N13268 applied on to of the current HEAD of 1_6_x.
[13:02:40] <wiesand> Obviously, 13268 will have to be merged on 1_8_x before it's available for 1.6.
[13:03:45] <wiesand> There are two more changes I'm wondering about:
[13:04:02] <wiesand> 13050 + 13051
[13:04:12] <wiesand> Are these for 1.6?
[13:04:16] <mvita> looking
[13:05:09] <mvita> I don't know offhand
[13:05:16] <kaduk@jabber.openafs.org/barnowl> The commit referenced in 13050's message does not exist on 1.6.x
[13:05:27] <wiesand> There's significant code skew in rx.c. Like 8296 never made it into 1.6, and even that's hard to backport for me.
[13:05:42] <wiesand> ben: right, but I think the problem does
[13:06:32] <wiesand> rxi_SendRawAbort hardcodes the serial number to 1
[13:06:42] <kaduk@jabber.openafs.org/barnowl> Ah, I see.
[13:08:07] <wiesand> but then, the commit message says it's just a help for humans reading packet traces…
[13:09:43] <kaduk@jabber.openafs.org/barnowl> Yeah, I don't think it's a particularly critical issue.
[13:10:01] <wiesand> ok, not for 1.6 at this late stage
[13:10:03] <wiesand> thanks
[13:10:05] <kaduk@jabber.openafs.org/barnowl> If someone made a patch, it would probably be easy enough to review,
and not harmful, but I'm not sure that we need to prioritize effort
into generating that patch
[13:10:21] <mvita> agreed
[13:11:53] <wiesand> so that's the 1.6 status - I have yet to deploy the test builds and actually test rpm building, but that should happen today
[13:12:11] <mvita> your list of candidates looks good and complete
[13:12:26] <wiesand> 13268 is kind of a blocker
[13:12:29] <mvita> I don't see anything else that should go in
[13:12:44] <kaduk@jabber.openafs.org/barnowl> I have it up in my browser, should be easy to review (but need to have
breakfast first)
[13:12:48] <wiesand> I'd love to see success reports from the 1.6 daily builders again
[13:12:55] <wiesand> ok, thanks
[13:13:09] <wiesand> mark: thanks too
[13:13:51] <wiesand> on to 1.8/master?
[13:13:58] <mvita> yes, it would be good to get 13268 merged for 1.8.x and then pull up for 1.6.23
[13:14:12] <mvita> 4.18 is due to be released on Sunday
[13:14:58] <wiesand> we may even have a 1.6.22.4 then
[13:15:07] <wiesand> and 1.8.1.1?
[13:15:15] <mvita> oh?
[13:15:20] <mvita> are you kidding?
[13:15:41] <wiesand> er, no…
[13:15:42] <kaduk@jabber.openafs.org/barnowl> We failed to get the time64 thing into 1.8.1, yes
[13:15:51] <mvita> ah.
[13:15:56] <kaduk@jabber.openafs.org/barnowl> OTOH, we did get 1.8.1 out the door
[13:16:06] <wiesand> rigth
[13:16:11] <mvita> I didn't realize you would consider a separate point release for that
[13:17:29] <kaduk@jabber.openafs.org/barnowl> I mean, point releases are basically just for new OS support
[13:17:30] <wiesand> both 1.6.23 and 1.8.2 will probably not make it out before Fedora & Co. update their kernels
[13:17:50] <mvita> yes
[13:19:22] <kaduk@jabber.openafs.org/barnowl> Not a whole lot to report from here.
[13:19:39] <wiesand> new 1.8.0 bug report in rt
[13:19:46] <mvita> As far as I know, we are only sure we _build_ on 4.18 - I have not tried to _run_ OpenAFS there, and I don't know if Joe has either.
[13:19:58] <kaduk@jabber.openafs.org/barnowl> 1.8.1 is out, no complaints yet.  It's in Debian, and hopefully I can
get it into FreeBSD ports as well (the first 1.8.x one, there)
[13:20:40] <wiesand> aren't the daily builders performing those robotests of Mike's?
[13:20:55] <mvita> Ben, I confirmed yesterday that 1.8.1 (and master) has a bug in rx_atomic that causes a kernel panic on Solaris SPARC
[13:21:05] <kaduk@jabber.openafs.org/barnowl> Yeah, I saw the thing in RT; pretty confusing for fs(1) to be
hung/deadlocked.
[13:21:24] <mvita> https://gerrit.openafs.org/#/c/12991/
[13:21:51] <mvita> I'm talking about a problem reported in SNA RT - I haven't seen the one in central RT
[13:21:55] <kaduk@jabber.openafs.org/barnowl> mvita: yeah, I saw that.  Unfortunately, the fix in gerrit nominally
depends on some changes that are not ready to merge yet
[13:22:15] <wiesand> AFS "Connection Timed Out" / fs checkservers hangs <https://rt.central.org/rt/Ticket/Display.html?id=134616>
[13:22:39] <kaduk@jabber.openafs.org/barnowl> mvita: so you need to yell at meffie to upload those fixed versions ;)
[13:22:58] <kaduk@jabber.openafs.org/barnowl> (or reorder the stack yourself, I suppose)
[13:23:32] <kaduk@jabber.openafs.org/barnowl> re RT134616, I froget how linux lets me look at the kernel call tree,
so I didn't reply yet
[13:24:55] <mvita> weisand:  no, the linux daily builder does not run the tests, it only builds
[13:25:18] <wiesand> Alt-SysRq-T ?
[13:25:38] <wiesand> mark: ah, pity
[13:26:44] <meffie> oh, someone is yelling at me again?
[13:28:02] <meffie> solaris fixes? yeah i yell at me too.
[13:29:14] <wiesand> btw, I tried billings' kafs stuff
[13:29:20] <kaduk@jabber.openafs.org/barnowl> In totally unrelated news, the TLS 1.3 RFC is very close to completed,
so I have some last-minute things to do there.
[13:30:42] <wiesand> re kafs: I didn't manage to obtain tokens, but that's probably due to our k5 setup
[13:31:00] <wiesand> but the client works like a charm
[13:31:01] <kaduk@jabber.openafs.org/barnowl> I'm waiting for Debian to get a 4.18 kernel, then I'll probably try it
out as well.
[13:31:24] <wiesand> I love seeing df tell the truth
[13:31:39] <wiesand> though it will probably break some dumb software again
[13:32:10] <kaduk@jabber.openafs.org/barnowl> No doubt
[13:33:21] <wiesand> maybe david could be convinced to make the client lie about free space in the afs root only?
[13:33:43] <kaduk@jabber.openafs.org/barnowl> David is generally reasonable
[13:34:27] <kaduk@jabber.openafs.org/barnowl> [13268 merged on 1.8.x]
[13:34:40] <wiesand> thanks, will pull it up right away
[13:35:15] <mvita> echo t > /proc/sysrq-trigger
[13:35:41] <meffie> ?
[13:35:46] <wiesand> or Alt-SysRq-T on the keyboard on a VC
[13:36:04] <mvita> Ben was asking how to dump kernel stacks on Linux
[13:36:19] <kaduk@jabber.openafs.org/barnowl> "send it to https://rt.central.org/rt/Ticket/Display.html?id=134616"
:)
[13:37:25] <meffie> can i give a buildbot progress report?
[13:37:30] <kaduk@jabber.openafs.org/barnowl> sure!
[13:38:33] <meffie> thanks ben for installing python3 (etc) on the server. i was able to make progress on the new buildbot master (finally)
[13:39:24] <meffie> the "plan" is to have a transition from the old version to the new version, and we now have both running on buildbot.openafs.org (on different ports).
[13:40:11] <meffie> the new one is running at: http://buildbot.openafs.org:8011 <http://buildbot.openafs.org:8011/#/>/
[13:40:55] <meffie> and the initial configuration and notes are in github: https://github.com/openafs-contrib/afsbotcfg   (branch buildbot-13x)
[13:41:26] <mvita> I updated the Ubuntu hang RT
[13:42:06] <meffie> it's just a basic config so far. the old master.cfg is not compatible with the buildbot version 1.x so i've decided to start fresh with a clean config.
[13:42:35] <mvita> ahh, Ben beat me to it  :-$
[13:42:40] <meffie> so far we just have one test worker and basic authorization for the web ui.
[13:42:54] <mvita> Sorry, I thought you asked me to update it.
[13:42:55] <meffie> i'll be adding a gerrit builder next
[13:43:40] <kaduk@jabber.openafs.org/barnowl> Yeah, I accept blame for the double reply
[13:43:56] <meffie> the ui is very different from the old buildbot, so it's going to take some time to learn the new system.
[13:44:11] <kaduk@jabber.openafs.org/barnowl> The old buildbot UI was ... not great, so that's probably okay.
[13:44:35] <meffie> yeah, the new ui is much nicer and is modern.
[13:45:52] <wiesand> 13268 pullup is 13269
[13:46:07] <meffie> also, i've setup a local instance for testing and making changes (based on the README.md in the afsbotcfg git repo on github)
[13:46:11] <wiesand> i'm really eager to merge that one… review review review
[13:46:32] <meffie> ok
[13:47:25] <wiesand> Is there more to discuss today?
[13:47:58] <kaduk@jabber.openafs.org/barnowl> Anything high-priority on master other than the solaris atomics fix?
[13:48:28] <mvita> there was another fix that prevents a problem on PPC
[13:48:35] <mvita> looking for details...
[13:49:28] <mvita> ah, yeah, that was for 1.6.x, and was already mentioned in Stephan's list
[13:49:45] <wiesand> which one?
[13:49:50] <mvita> https://gerrit.openafs.org/13261 LINUX: Don't compile syscall code with keyrings
[13:50:01] <kaduk@jabber.openafs.org/barnowl> Ah, right, that one
[13:50:20] <kaduk@jabber.openafs.org/barnowl> For 1.6, even(!)
[13:51:09] <wiesand> objections?
[13:51:13] <meffie> > other than the solaris atomics fix?
which gerrit is that?
[13:51:21] <mvita> that's two fixes this week that were ostensibly for hygeine but actually repaired issues on "exotic" architectures
[13:52:12] <mvita> https://gerrit.openafs.org/#/c/12991/
[13:52:17] <mvita> atomics
[13:52:23] <meffie> ok, thanks
[13:52:33] <kaduk@jabber.openafs.org/barnowl> actual atomics fix are https://gerrit.openafs.org/#/c/12991/1
your dependencies are https://gerrit.openafs.org/#/c/12961/3 and
https://gerrit.openafs.org/#/c/12956/4 IIRC
[13:52:40] <mvita> Your "warning" fix is actually required to prevent a panic on SPARC
[13:52:56] <meffie> well, we were warned :)
[13:53:02] <mvita> zackly
[13:53:07] <kaduk@jabber.openafs.org/barnowl> Funny how warnings work like that :)
[13:53:09] <meffie> it's not a good idea to ignore them :)
[13:53:49] <wiesand> and i believed they're just noise…
[13:54:09] <meffie> some are, but some maybe not.
[13:54:37] <kaduk@jabber.openafs.org/barnowl> If you want to see some noise, try clang -Weverything :)
[13:54:46] <meffie> oh man.
[13:54:56] <mvita> -Wkitchensink
[13:55:11] <meffie> ok, need to make solaris-warnings stack a priority.
[13:55:39] <meffie> i was having so much fun with the buildbot. nice new clean installation and config :)
[13:55:40] <mvita> well, at least the one that had some bite to it
[13:56:28] <meffie> (also on the buildbot front we added a solaris 11 sparc builder)
[13:56:37] <mvita> mike is our 10o'clock postponed?
[13:57:02] <meffie> yes
[13:57:23] <mvita> SPARC BUILDER is quite slow - unless you use all 32 cores in a build (as I did last night)
[13:57:35] <mvita> I wish I could have heard the fans spin up
[13:57:41] <wiesand> ot: my good old white macbook finally broke yesterday :-(
[13:58:00] <kaduk@jabber.openafs.org/barnowl> :(
[13:58:17] <meffie> :(
[13:58:33] <mvita> <moment of silence for the macbook>
[13:59:06] <wiesand> thank you…
[13:59:17] <wiesand> ok, next try: anything else to discuss today?
[13:59:28] <mvita> nope.
[13:59:39] <meffie> have a good weekend.
[13:59:48] <kaduk@jabber.openafs.org/barnowl> You, too!
[14:00:02] <wiesand> Let's adjourn. Thanks a lot everybody!
[14:00:30] wiesand leaves the room
[14:18:16] meffie leaves the room
[14:44:59] mvita leaves the room
[15:24:42] mvita joins the room
[15:26:19] <mvita> Joe has built and tested master on 4.18-rc7 - all is well
[15:26:53] <kaduk@jabber.openafs.org/barnowl> Yay!
[15:42:23] mvita leaves the room
[17:28:21] mvita joins the room
[22:58:49] mvita leaves the room