Home
release-team@conference.openafs.org
Friday, January 18, 2019< ^ >
Room Configuration
Room Occupants

GMT+0
[13:59:23] wiesand joins the room
[14:00:26] mvita joins the room
[14:00:55] <wiesand> Good morning
[14:01:00] <mvita> good morning
[14:01:37] <mvita> I don't see Mike online yet
[14:01:49] <wiesand> nor do I
[14:01:55] <kaduk@jabber.openafs.org/barnowl> greetings
[14:02:11] <wiesand> Mark: hope you're better?
[14:02:14] <wiesand> Hello Ben
[14:02:18] <mvita> yes
[14:02:34] <mvita> although I'm more congested this morning again
[14:02:41] meffie joins the room
[14:02:42] <mvita> I can work
[14:02:48] <meffie> good morning
[14:02:53] <mvita> meffie!
[14:02:54] <wiesand> Hello Mike
[14:03:15] <meffie> apologies for lateness
[14:03:18] <wiesand> I have to confess that another week passed by w/o much activity on my part :-(
[14:04:11] <wiesand> I added the SRPMs for the 1.6.23/1.8.2 releases though, as discussed last week.
[14:04:17] <meffie> same here, but andrew mark and cheyenne have been busy
[14:04:33] <meffie> nice, thank you.
[14:04:39] <kaduk@jabber.openafs.org/barnowl> Yeah, lots of activity in gerrit for me to try to catch up on
[14:05:07] <mvita> no, I've been tied up with a lot of non-work stuff
[14:05:43] <wiesand> Same here, and I'll be very busy for another week.
[14:06:20] <wiesand> But I noticed the gerrit activity recently. In particular, we seem to have chnages for Linux 5.0 support. Great!
[14:06:33] <wiesand> Are these believed to be complete?
[14:07:02] <kaduk@jabber.openafs.org/barnowl> The first two IESG telechats of the year are pretty light workloads,
but a lot of stuff backed up while I was on holiday and I've mostly
been trying to catch up on that
[14:07:03] <mvita> yes
[14:07:05] <meffie> yes, i think so. i helped cheyenne test them yesterday
[14:07:43] <kaduk@jabber.openafs.org/barnowl> The mail notificattion made them sound pretty straightforward, but I
haven't opened them up to look yet
[14:08:34] <meffie> yes i think they are straightforward
[14:08:41] <wiesand> These would probably make sense to include in the prereleases. I'd appreciate it if they could get priority on master ;-)
[14:08:52] <mvita> ok
[14:08:56] <kaduk@jabber.openafs.org/barnowl> ok
[14:08:57] <meffie> ok
[14:09:40] <wiesand> Any other changes of plans for 1.6.24/1.8.3?
[14:10:09] <meffie> andrew pushed a fix for a rather serious bug
[14:10:14] <wiesand> (besides more delay to be blamed on me?)
[14:10:18] <mvita> yes, it's a lulu
[14:10:25] <wiesand> lulu?
[14:10:33] <meffie> a big one
[14:10:51] <wiesand> Andrew pushed lots of fixes recently… which on'e the big one?
[14:10:57] <kaduk@jabber.openafs.org/barnowl> zero-length chunks?
[14:11:08] <meffie> topic: libafs-enospace-corruption
[14:11:22] <mvita> https://www.merriam-webster.com/dictionary/lulu
[14:12:25] <meffie> this was found after an incident
[14:13:32] <meffie> basically a full cache partition could lead to file curruption
[14:14:37] <meffie> the critical gerrit is 13428
[14:15:15] <wiesand> will this corrupt files on the server, or only on the affected client?
[14:15:27] <meffie> and the bug is present in all 1.6+ versions
[14:15:52] <meffie> hmm, i dont know offhand...
[14:16:26] <mvita> client only
[14:16:28] <wiesand> looks like the latter to me, but…
[14:17:26] <wiesand> ok, that makes it not quite as severe
[14:17:53] <wiesand> and it can't be the reason for my "idledead foo"
[14:18:29] <mvita> ???
[14:20:19] <kaduk@jabber.openafs.org/barnowl> mvita: Stephan is (rightfully) paranoid about things that cause data
corruption on the server, with idledead being the poster boy
[14:20:38] <wiesand> RT #133023
[14:21:07] <wiesand> Daria always claimed ideldead isn't the culprit, just the trigger
[14:22:22] <wiesand> A bug similar to this one in the write code path could explain that
[14:22:27] <mvita> "no permission to view ticket" sigh
[14:22:49] <kaduk@jabber.openafs.org/barnowl> Send me mail reminding me to yell at jhutz about RT permissions
[14:23:09] <wiesand> Well it's in the security queue
[14:23:31] <kaduk@jabber.openafs.org/barnowl> mvita: try again (force-reload if needed)
[14:23:51] <mvita> got it - thanks
[14:27:20] <meffie> do we want to include 13428 and friends in 1.6.24pre1 ?
[14:27:22] <wiesand> so yes, 13428 looks like an important fix, but the issue documented in #133023 is what I would call a lulu (thanks for the pointer)
[14:28:10] <wiesand> I wouldn't want to block on it, but if it's ready in time, why not??
[14:29:08] <meffie> ok, do you have a cutoff time?
[14:29:44] <meffie> i think we need to do some reviews for master first.
[14:30:00] <wiesand> "whenever I finally get around to seriously work on the pre1s" - couple of days at least
[14:30:16] <meffie> ok
[14:30:38] <wiesand> yes, of course, this should go through the usual "review thoroughly, merge on master, pull up, review again,…" process
[14:31:07] <mvita> I think 13428 should take priority over the Linux5 stuff, since we have plenty of time for the latter
[14:31:51] <mvita> Sunday will only be -rc3
[14:32:44] <meffie> it would be nice to have them in for the buildbot
[14:32:52] <mvita> true
[14:33:37] <wiesand> My worry regarding 13428 is: could this introduce data corruption on clients just because a file server is very busy?
[14:35:14] <kaduk@jabber.openafs.org/barnowl> I'll keep that in mind when reviewing
[14:35:27] <meffie> this = the bug?
[14:35:31] <mvita> I don't know - but my current understanding is that this is only triggered by cache issues (esp. cache full)
[14:35:49] <wiesand> this = the fix
[14:36:01] <mvita> I'm still reviewing it too, and I also have my own development stack that is somewhat related
[14:36:48] <wiesand> The commit message mentions that "Theoretically this can be seen from an
unexpected error from Rx, […]"
[14:37:16] <mvita> yes, but then he goes on to say that would be handled correctly by later processing
[14:38:08] <mvita> also, for the purposes of this issue, I think "busy" doesn't count as an rx error
[14:38:19] <wiesand> Sorry, it's just the paranoia the idledead episode left me with
[14:38:36] <mvita> still worth studying and verifying, it's a good question
[14:39:17] <meffie> wiesand really doe not like idledead
[14:39:36] <mvita> now that I finally have the backstory, I understand
[14:39:43] <mvita> it's completely justified
[14:40:04] <meffie> i agree, just saying
[14:40:10] <mvita> holes are for donuts, not data
[14:41:21] <wiesand> Thanks for your understanding :) But again: yes, looks like an important fix.
[14:41:45] <kaduk@jabber.openafs.org/barnowl> (I left an "I am an idiot; please help" comment on 13433 if anyone has
state at hand on the linux 5.0 stuff)
[14:42:17] <meffie> ok
[14:42:40] <mvita> looking
[14:44:13] <mvita> I'll try to take a look later today, but Cheyenne will probably get there first
[14:44:24] <mvita> re: include chain
[14:46:41] <meffie> btw, cheyenne is a big gentoo fan, so he likes to compile, all the time.
[14:46:57] <mvita> oh yeah
[14:48:29] <meffie> (the openafs gentoo maintainer is inactive so he may help there too)
[14:49:10] <wiesand> ok, looks like we have a slightly more grown wish list for the prereleases, but no substantial change of plans
[14:49:20] <meffie> yes
[14:50:13] <wiesand> Are there more items to discuss regarding the stable series?
[14:50:29] <meffie> none from i
[14:50:37] <mvita> nor here
[14:51:09] <wiesand> On to the exciting stuff then? (master, future release, …)?
[14:52:09] <meffie> i'm working on an update for pthreaded-xtat for master
[14:52:42] <meffie> to remove more lwp
[14:53:14] <meffie> thanks ben for the comments
[14:53:24] <meffie> (i struggle with our makefiles)
[14:56:09] <kaduk@jabber.openafs.org/barnowl> Sorry I did not have more comments -- I only got about halfway through
the review so far
[14:57:43] <meffie> reviews are very time consuming, thank you.
[14:59:08] <meffie> anything else Ben?
[14:59:47] <kaduk@jabber.openafs.org/barnowl> Um, just a sec
[15:01:12] <kaduk@jabber.openafs.org/barnowl> Hmm, I'm not finding anything to help me remember whatever it was I
was trying to think of.
[15:01:27] <meffie> heh. been there.
[15:01:39] <wiesand> me too
[15:01:57] <kaduk@jabber.openafs.org/barnowl> I guess in some of the changes I've looked at recently they already
had a +1 from me with some comment about "wait for <someone> to chime
in" or similar; I don't have a great way to remind myself to check
back a week later, so it may be worth pinging me if you run into such
things.
[15:02:12] <meffie> random question. is there any reason we still have klog.krb5 ? is it still useful?
[15:02:22] <wiesand> oh yes
[15:03:14] <kaduk@jabber.openafs.org/barnowl> Yeah, some sites really like the klog semantics
[15:03:21] <wiesand> I use it every time I upload to g.c.o.
[15:04:04] <wiesand> It is misnamed though. Should be called "k5log" IMO ;-)
[15:06:09] <meffie> ah, yes. ok, thanks.
[15:07:02] <kaduk@jabber.openafs.org/barnowl> I don't think I have anything else for today, if we're wrapping up
[15:07:24] <meffie> ok, thanks have a good weekend.
[15:07:32] <wiesand> Let's call it a meeting then. Thanks a lot everybodey!
[15:08:21] wiesand leaves the room
[15:08:48] meffie leaves the room
[15:09:58] <kaduk@jabber.openafs.org/barnowl> Yes, thanks everyone!
[15:53:44] meffie joins the room
[16:32:35] mvita leaves the room
[16:36:56] meffie leaves the room
[18:05:55] meffie joins the room
[19:29:00] meffie leaves the room
[20:44:08] mvita joins the room