- Friday, February 23, 2018

release-team@conference.openafs.org

Friday, February 23, 2018< ^ >

Room Configuration

Room Occupants

GMT+0
[00:52:57] mvita leaves the room
[03:08:58] mvita joins the room
[04:18:25] mvita leaves the room
[11:15:28] kaduk@jabber.openafs.org/barnowl leaves the room: Replaced by new connection
[11:15:28] kaduk@jabber.openafs.org/barnowl joins the room
[11:56:44] Marcio Barbosa joins the room
[13:55:12] wiesand joins the room
[13:55:12] kaduk@jabber.openafs.org/barnowl leaves the room
[13:56:46] meffie joins the room
[13:57:45] kaduk@jabber.openafs.org/barnowl joins the room
[14:00:10] <wiesand> Good morning
[14:00:27] <kaduk@jabber.openafs.org/barnowl> Great, looks like I can receive but not send (!)
[14:00:42] <Marcio Barbosa> good morning
[14:00:55] <kaduk@jabber.openafs.org/barnowl> Huh, second time's the charm, or something.
[14:01:35] <meffie> good morning!
[14:01:48] <wiesand> Sorry for not having gotten much done this week :-(
[14:02:13] <wiesand> I did the ubik rebases and a few more pullups today.
[14:02:31] <wiesand> But buildbot seems to be having trouble. Haven't checked yet.
[14:02:33] <kaduk@jabber.openafs.org/barnowl> Slow progress is still progress
[14:04:00] <wiesand> Is 12914 for 1.6?
[14:04:05] <meffie> thank you wiesand!
[14:04:25] <kaduk@jabber.openafs.org/barnowl> Might as well take 12914, yeah
[14:05:18] <wiesand> There's some code skew there… any volunteers for pulling it up to 1.6?
[14:05:28] <meffie> i can do it.
[14:05:30] <kaduk@jabber.openafs.org/barnowl> Oh, and while I remember -- I'll need to duck out for ~10 minutes at
some point, unfortunately.
[14:05:34] <wiesand> Thanks Mike
[14:05:50] <meffie> if i get into trouble, i'll ask for help :)
[14:06:31] <wiesand> I think I don't have much more on 1.6 today. It's still mostly blocked on me doing my homework.
[14:07:06] <kaduk@jabber.openafs.org/barnowl> We don't have a rush due to upstream stuff yet, though, right?
[14:07:15] <kaduk@jabber.openafs.org/barnowl> Just the RHEL7.5 thing?
[14:07:29] <wiesand> Yes, just that so far.
[14:07:49] <wiesand> May need a branch for a 1.6.22.3 …
[14:08:03] <wiesand> That is, if we manage to fix it…
[14:08:10] <meffie> sorry i dont have much to report on that front yet. mark has been out with the flu (it's a bad flu season)
[14:08:28] <kaduk@jabber.openafs.org/barnowl> Okay, moving on to 1.8, then, I guess.
The Debian builds of 1.8.0pre5 all failed (except for my local one)
with some type error about u_short not recognized, but I haven't
gotten a chance to look yet.
[14:08:36] <wiesand> same here, and that's partially an excuse for my slowness too
[14:08:42] <kaduk@jabber.openafs.org/barnowl> Yeah, I've heard of several people with bad flu experiences.
[14:08:53] <kaduk@jabber.openafs.org/barnowl> Hopefully everyone can get well soon and/or stay healthy, as
appropriate.
[14:10:41] <kaduk@jabber.openafs.org/barnowl> I don't know if this u_short thing is more fallout from the new glibc
or what (I didn't try another local build yet because I was considering
saving the build artifacts from the previous one, i.e., the one I
actually uploaded).
[14:11:32] <meffie> btw, which debian are you using, 9.3?
[14:11:53] <kaduk@jabber.openafs.org/barnowl> We also got this report on the list about some build warnings/errors
with FreeBSD 11.1, which I can reproduce locally.  But, it's only in
the kernel module build, which I'm pretty sure is going to fail anyway
due to upstream changes, so once I can confirm that (by hacking things
to get past the posted error) I'll reply back to the user.
[14:12:02] <kaduk@jabber.openafs.org/barnowl> These are for debian unstable :))
[14:12:09] <meffie> oh, ok.
[14:12:53] <kaduk@jabber.openafs.org/barnowl> The way debian works means that you basically don't get a new upstream
version into an actual extant debian release.
[14:14:03] <meffie> ok, right.
[14:15:36] <kaduk@jabber.openafs.org/barnowl> (I just realized that the -Werror is coming from upstream's build
system, since we are using that now instead of our own pale imitation
as we do on 1.6)
[14:16:16] <kaduk@jabber.openafs.org/barnowl> Anyway, if my hypothesis is correct, no code changes will be needed
for 1.8 final in this regard.
[14:17:05] <kaduk@jabber.openafs.org/barnowl> I did merge a couple changes -- a documentation fix and the
nul-termination for rx version one
[14:17:31] <meffie> and the rpm libuafs stuff
[14:17:39] <kaduk@jabber.openafs.org/barnowl> But I don't think that those would require a new release candidate
before final release.  (Whatever we end up doing for the debian build
issue might, though.)
[14:17:53] <wiesand> +1
[14:17:55] <kaduk@jabber.openafs.org/barnowl> Oh, right, I did that, too -- thanks again for putting those patches
together :)
[14:18:22] <meffie> sorry i missed that before pre5 :(
[14:18:44] <wiesand> that's not a 1.6 issue, right?
[14:18:44] <kaduk@jabber.openafs.org/barnowl> It's okay.
[14:18:50] <meffie> correct
[14:19:07] <meffie> i'm pretty sure. i'll double check.
[14:20:28] <kaduk@jabber.openafs.org/barnowl> I thought we had some long-standing issue with the swig/ukernel stuff
on 1.6 for a long time until we finally implemented a configure
opt-out.
[14:20:52] <kaduk@jabber.openafs.org/barnowl> ... but that override is only on 1.8, never mind
[14:21:03] <kaduk@jabber.openafs.org/barnowl> (12518, FTR)
[14:22:21] <meffie> > that's not a 1.6 issue, right?
yes, that's 1.8.x only.
[14:23:27] <kaduk@jabber.openafs.org/barnowl> Anyway, so 1.8 is still looking in pretty good shape; not much more to
say there that I remember.
[14:23:45] <wiesand> The F20 builder is offline again. I'll ping Derek.
[14:25:04] <kaduk@jabber.openafs.org/barnowl> Looking at master ... thanks for the venus/butc string-handling stuff,
Mike.  Any thoughts about squashing down both vs just butc (or pushing
back against my heavy hand)?
[14:26:13] <meffie> according to wikipedia, fedora 20 eol was back in 2015, should it be downgraded to a nightly builder?
[14:26:34] <kaduk@jabber.openafs.org/barnowl> That would seem to make sense
[14:27:16] <wiesand> the list is fairly short anyway… looks like the opensuse12 i386 builder was discontinued lately too?
[14:28:18] <meffie> looking at the config, we have a lot of fedora builders, but only f20 is a gerrit builder!?
[14:28:29] <meffie> for 1.6.x
[14:28:41] <kaduk@jabber.openafs.org/barnowl> opensuse 12.3 was released in 2013 and is the last one from that series
listed in the wikipedia article
[14:29:02] <kaduk@jabber.openafs.org/barnowl> I thought we were going to promote more of the fedorao builders to
triggered.  Maybe that only happened for master
[14:30:04] <kaduk@jabber.openafs.org/barnowl> Anyway, Mike, did you miss my question about the string-handling
changes while looking at buildbot stuff?
[14:30:21] <meffie> i think that's right, the master list has many, many more gerrit builders. should i update the config?
[14:31:05] <kaduk@jabber.openafs.org/barnowl> Yeah, I would say go ahead and update it
[14:31:08] <wiesand> please do, but maybe omit the very slow ones (the windiws ones i believe)
[14:31:24] <meffie> ok, wiesand, i will do that. thanks.
[14:31:32] <wiesand> thank you
[14:32:06] <meffie> Ben, i pushed to new gerrits yesterday. did i miss a question?
[14:32:34] <kaduk@jabber.openafs.org/barnowl> I asked one in the room here -- thoughts about squashing one or both
of them (venus, butc)?
[14:33:37] <kaduk@jabber.openafs.org/barnowl> Oh, and backing up a bit I can report that my freebsd 11.1 build with
-Wno-pointer-sign in MakefileProto.FBSD did complete, to my surprise.
Maybe the issues only show up at runtime :-/
[14:34:52] <meffie> oh, i see the questions on 12922, 12923; yes, i can squash them if you think that is better. i wasnt sure, so i did them as separate commits because it is easier to squash than to de-squash.
[14:35:06] <kaduk@jabber.openafs.org/barnowl> Definitely agreed on easier to squash than to de-squash
[14:35:41] <kaduk@jabber.openafs.org/barnowl> I was only pro-squashing for the butc one because the array that you
make bigger in the first commit disappears entirely in the second one,
so there's not a whole lot of point in keeping the first commit
around.
[14:36:10] <kaduk@jabber.openafs.org/barnowl> For the venus stuff we don't switch to asprintf, so the first commit
is a bugfix and the second one general cleanup.  Or maybe this
distinction only makes sense inside my head
[14:36:44] <wiesand> no guaranteed asprintf on 1.6…
[14:38:53] <kaduk@jabber.openafs.org/barnowl> Hmm, that's a good point.  So you'd want to take the bugfixes for 1.6
but skip the general cleanup?
[14:38:58] <meffie> ah, good point wiesand!
[14:39:23] <meffie> now i'm glad i made them separate :)
[14:40:26] <meffie> these are to fix building on suse tumbleweed
[14:40:57] <kaduk@jabber.openafs.org/barnowl> I may have to pop out soon, so let me mention what I think is the last
topic I had: Marcio came up with 12924, which looks like a fine change
to have.  Did we hear back from Hartmut about what caused his
particular error case?
I am also unsure that taking 12924 (which I think we should do) fully
eliminates the "need" (not necessarily actual need) to take something
else as well, like 12905 but probably somewhat different.
[14:41:43] <Marcio Barbosa> > Did we hear back from Hartmut about what caused his
particular error case?
[14:41:44] <meffie> yes i did hear from harmut, i'll send mail. the bug was triggerred by rx/osd patches he was doing
[14:41:57] <Marcio Barbosa> right
[14:42:12] <Marcio Barbosa> he had two sync-sites at the same time.
[14:42:25] <kaduk@jabber.openafs.org/barnowl> two sync-sites at the same time, that sounds exciting.
[14:42:31] <meffie> yes
[14:42:48] <Marcio Barbosa> but it seems he changed the code to allow the election even when two sites are down (he has 3 sites)
[14:43:30] <Marcio Barbosa> i would expect problems
[14:45:07] <wiesand> sounds not quite split brain safe
[14:45:12] <meffie> for non-patched code, i think the only way to trigger this bug is to have a misconfigured cellservdb, but still that does happen far more than we want.
[14:46:17] <kaduk@jabber.openafs.org/barnowl> misconfigured being different across sites?
[14:46:20] <kaduk@jabber.openafs.org/barnowl> GONE
[14:46:23] <meffie> yes
[14:47:27] <wiesand> derek restarted the f20 build slave - it's fast so should catch up soon
[14:48:10] <wiesand> Looks like Ben left us… anything else to discuss?
[14:48:45] <meffie> i dont have anything else.
[14:49:04] <Marcio Barbosa> can we talk about 12283 and 12290 (very briefly)?
[14:49:12] <meffie> oh, yes.
[14:49:21] <Marcio Barbosa> i am not sure if these patches are necessary.
[14:49:35] <Marcio Barbosa> 12283 - what is the point of setting ‘rcode’ to RX_CALL_DEAD and then return UNOQUORUM to the caller? 12289 already returns UNOQUORUM if ‘rcode == 0’.
[14:49:49] <Marcio Barbosa> 12290 - we return the original code to the caller, unless ‘rcode’ is equal to 0 or RX_CALL_DEAD. i am not sure if that log would be helpful.
[14:50:01] <Marcio Barbosa> in my option, these patches are not really necessary. 12289 is enough.
[14:51:49] <wiesand> Can you add this analysis as comments to the gerrits?
[14:51:59] <Marcio Barbosa> sure
[14:52:47] <wiesand> It'd be fine if you're right here.
[14:53:16] mvita joins the room
[14:53:33] <Marcio Barbosa> cool. i will update the gerrits
[14:53:33] <wiesand> Welcome Mark
[14:53:59] <meffie> if so, it would be good if andrew agrees, that he mark them abondoned, with the reasons.
[14:54:27] <wiesand> or not ;-) (that's the idea)
[14:54:41] <mvita> hi
[14:55:46] <wiesand> Hi - a bit late since we're just done, but maybe you bring some news?
[14:56:25] <mvita> no, I have nothing, I've been out sick most of the week
[14:56:30] <mvita> and still am, really
[14:57:00] <meffie> oh dear, get well soon mvita
[14:57:21] <wiesand> get well soon (i hope your furnace is back in good shape?)
[14:57:42] <mvita> yup furnace is humming right along!
[14:58:01] <wiesand> at least that
[14:58:23] <wiesand> But it seems we're finished for today? Adjounr?
[14:59:28] <meffie> thank you, have a good weekend.
[15:00:00] <wiesand> Thanks a lot everyone. I'll try to do better than last week, I promise.
[15:00:15] wiesand leaves the room
[15:17:10] meffie leaves the room
[15:46:58] mvita leaves the room
[19:56:18] mvita joins the room
[20:50:06] mvita leaves the room
[20:53:05] mvita joins the room
[21:43:22] Marcio Barbosa leaves the room