- Friday, May 25, 2018

release-team@conference.openafs.org

Friday, May 25, 2018< ^ >

Room Configuration

Room Occupants

GMT+0
[00:53:23] mvita joins the room
[01:45:45] mvita leaves the room
[12:37:00] kaduk@jabber.openafs.org/barnowl leaves the room
[12:37:51] kaduk@jabber.openafs.org/barnowl joins the room
[12:44:10] <kaduk@jabber.openafs.org/barnowl> third try...jabber is working today, right?
[12:51:56] meffie joins the room
[12:52:52] <meffie> good morning. network may be spotty for me
[12:53:59] <kaduk@jabber.openafs.org/barnowl> greetings
[12:56:55] <meffie> greetings
[12:57:55] <meffie> i was happy to see contributions from Pat. hope we didnt scare him away yet.
[12:59:41] <kaduk@jabber.openafs.org/barnowl> agreed on all counts
[13:00:43] wiesand joins the room
[13:02:22] <meffie> good afternoon wiesand
[13:02:41] <wiesand> Ah here we are. Hello
[13:03:42] mvita joins the room
[13:03:53] <mvita> good morning
[13:03:57] <wiesand> I made an attempt to pull up the ctf thing: 13083
[13:04:02] <mvita> Mike!
[13:04:17] <wiesand> "!" ?
[13:04:31] <mvita> I thought he was not attending today
[13:04:51] <mvita> I think it may have been a trick to get me here on time....
[13:05:01] <kaduk@jabber.openafs.org/barnowl> Hi Mark.  You should hopefully be getting mail from gerrit with a few
things that you may (or may not) want to update patchsets for.
[13:05:03] <mvita> ;-)
[13:05:13] <mvita> okey
[13:05:41] meffie leaves the room
[13:06:36] <wiesand> Except for 12083, no progress on my side in the past week :-(
[13:07:23] <wiesand> well, I'm probably an expert for chlorinated carbohydrates in ground water now…
[13:08:44] <wiesand> Things should return to normal "soon" here though
[13:09:08] <kaduk@jabber.openafs.org/barnowl> I hope the excitement has not caused too much stress for you.
[13:09:08] <wiesand> I see there's quite a bit of churn on master/1.8 - is anything of that headed for 1.6.x?
[13:09:53] <wiesand> Ben: no, actually it's quite fun
[13:09:56] <mvita> looking…
[13:10:03] meffie joins the room
[13:10:37] <meffie> (sorry, in a car.)
[13:10:40] <kaduk@jabber.openafs.org/barnowl> I don't remember much going by that would be needed on 1.6, but I also
only skimmed a lot of the gerrit mail.
[13:10:59] <mvita> it depends - now that 1.8.x is the new stable, does that mean 1.6.x gets (important) bug fixes only?
[13:11:20] <kaduk@jabber.openafs.org/barnowl> I think that was basically the intent, yeah.
[13:12:00] <meffie> the getdslot-eintr topic could be a 1.6,x candidate
[13:12:13] <kaduk@jabber.openafs.org/barnowl> (not sure if I'm jumping off-agenda, but) do we know if Andrew has an
ETA for updates on, e.g., getdslot-eintr?
[13:12:20] <mvita> for instance, the ubik_logging stuff would be nice to have in 1.6.x, but they are not technically addressing bugs - just making existing problems easier to debug
[13:12:22] <wiesand> MikeL yes, you mentioned that
[13:12:59] <kaduk@jabber.openafs.org/barnowl> Hmm, we have ubik_logging and ubik-log-stdio topics.  Anyone know
offhand which order is best to take them in?
[13:13:09] <wiesand> Mark: 1) we should relax that rule a bit until we're convinced that 1.8 is really stable and rather bug free
[13:14:12] <wiesand> 2) really low risk changes are still fine for the time being, especially if they help avoid skew in busy areas of the code
[13:14:31] <mvita> ubik-log-stdio is for the internal DB log, not the external server log like ubik-logging
[13:14:37] <mvita> so they are orthogonal
[13:14:49] <mvita> or should I say orth-log-onal
[13:15:07] <kaduk@jabber.openafs.org/barnowl> pun++
thanks
[13:15:07] <meffie> heh
[13:15:52] <wiesand> well, they aren't orthogonal from gerrit's point of view...
[13:16:01] <mvita> however, there is another patchset that changes all the ubik_*print* to ViceLog calls
[13:16:13] <mvita> that one REALLY conflicts with ubik-logging
[13:16:25] <kaduk@jabber.openafs.org/barnowl> Ah, right, in the ubik_perf topic.
[13:16:35] <mvita> aye.
[13:16:54] <kaduk@jabber.openafs.org/barnowl> "So who's on first?"
[13:17:12] <mvita> that is indeed a perf impact - but maybe for merge management's sake the logging changes in ubik-perf should be split to a separate topic
[13:17:28] <wiesand> looks like ubik-log-stdio sits on top of the ubik-loggin stack?
[13:17:53] <kaduk@jabber.openafs.org/barnowl> But maybe we care about the ubik_perf stack as higher priority than
ubik_logging?
[13:17:57] <mvita> or perhaps I can provide a single topic stack that includes both the ubik_perf logging changes and ubik-logging
[13:18:07] <mvita> that would be best for handling merge to master
[13:18:26] <wiesand> ah, no, that's something different again..
[13:18:47] <kaduk@jabber.openafs.org/barnowl> If you're up for making the consolidated stack, that seems helpful.
[13:19:21] <kaduk@jabber.openafs.org/barnowl> Should I hold off on reviewing/merging "convert ubik_print to ViceLog"
then?
[13:19:32] <mvita> if all these are indeed going to be merged, someone's got to handle the conflicts anyway - it would be best to do those in the submissions themselves
[13:20:31] <mvita> kaduk:  no, I think it would be helpful to get your input on those
[13:21:10] <kaduk@jabber.openafs.org/barnowl> Okay.
[13:21:15] <mvita> re: priorities
[13:22:05] <mvita> I guess large sites would benefit most from ubik-perf
[13:22:32] <mvita> typically DB servers don't carry a lot of load
[13:22:33] <kaduk@jabber.openafs.org/barnowl> *nods*
[13:23:04] <mvita> the logging benefits any site that suffers from quorum instabilities
[13:23:30] <mvita> which may be rare, but when they occur you need answers quick
[13:23:40] <kaduk@jabber.openafs.org/barnowl> true
[13:27:17] <mvita> as far as stability goes, the ubik-logging changes (and the perf logging changes) are very uncontroversial and should be completely safe for stable releases - I would like to see them in 1.6.x
[13:28:59] <mvita> some of the perf fixes may require more scrutiny due to their nature.  Although I'd like to see all those in 1.6.x as well, I can understand any reluctance for that.
[13:29:14] <wiesand> no objection here - we should tighten the rules for what goes in and what not gradually, as 1.8 stabilizes
[13:29:30] <wiesand> (re logging)
[13:29:58] <wiesand> let's discuss perf fixes after they're merged on 1.8
[13:30:10] <wiesand> but yes, expect some reluctance…
[13:30:14] <mvita> that sounds very resasonable
[13:31:38] <wiesand> we have reports from two sites about 1.8 clients failing to notice volume releases - any ideas?
[13:32:07] <mvita> well…
[13:32:08] <kaduk@jabber.openafs.org/barnowl> That was my next topic -- thanks for bringing it up!
(I have no ideas yet.)
[13:32:57] <mvita> we don't have callbacks for the volume lookup cache on the client - they time out (eventually)
[13:33:28] <mvita> so you would normally need to flush that to pick up a new site (or a site removal)
[13:34:04] <mvita> but that doesn't seem to be a factor here?  I haven't looked closely yet, but meffie and I were taking a quick look yesterday
[13:34:54] <meffie> no ideas yet. sorry.
[13:35:49] <meffie> (btw, the volume cache effectively does not time out, which is why we added the --volume-ttl option)
[13:36:05] <kaduk@jabber.openafs.org/barnowl> (Yeah, thanks for --volume-ttl)
[13:41:12] <wiesand> More to discuss today?
[13:41:59] <kaduk@jabber.openafs.org/barnowl> Should we hold 1.8.1 until we have a fix for the volume-release thing?
[13:42:51] <mvita> it needs to be root-caused ASAP.  Do we have anything urgent already scheduled for 1.8.1?
[13:43:43] <mvita> I see only the packaging stuff so far….
[13:44:17] <meffie> there's the crash when cache bypass is enabled.
[13:44:55] <mvita> ah, okay
[13:44:56] <kaduk@jabber.openafs.org/barnowl> "Make the public API usable", cache bypass, and the BSD dcache free
thing, are the main 1.8.1 things I remebmer
[13:45:09] <kaduk@jabber.openafs.org/barnowl> s/dcache/req/, whoops
[13:45:32] <mvita> then there's this sleeper which is actually pretty crucial:  http://git.openafs.org/?p=openafs.git;a=commit;h=b41065f2b8877580a7e1858b8e2857973ddf6503
[13:46:30] <mvita> I wouldn't hold 1.8.1 for the volume release thing, since we don't even have a handle on it yet.
[13:46:32] <meffie> and the macos fixes, but "no one" is running 1.8.x on macos yet
[13:46:43] <mvita> I am ;-)
[13:46:55] <mvita> are you saying I'm a nobody?
[13:47:23] <meffie> no one that is not sending patches :)
[13:47:46] <mvita> darn skippy
[13:50:10] <wiesand> I can't even find the OS X changes - looks they already drowned in all the new stuff
[13:50:50] <wiesand> Constantly dumping new code into gerrit much faster than it can be consumed is not going to be helpful…
[13:51:30] <wiesand> (IMHO)
[13:52:18] <mvita> Tell my boss ;-)
[13:52:49] <wiesand> Ok, will do
[13:53:38] <mvita> revealed on OS X but not OS X specific:   https://gerrit.openafs.org/#/c/13075/
[13:53:39] <meffie> ok, switching to github in 3, 2, ...
[13:54:05] <kaduk@jabber.openafs.org/barnowl> "So when do we get a new patchset on that one?" :-P
[13:54:48] <meffie> i think marcio found some problems on macos (build issues), yet to be pushed to gerrit.
[13:55:22] <kaduk@jabber.openafs.org/barnowl> I think there are already some OS X bits from Marcio in gerrit ...
unless I merged them and forgot about it?
[13:55:53] <mvita> > "when…"
[13:56:19] <mvita> Constantly dumping new reviews into gerrit much faster than they can be addressed is not going to be helpful…
[13:56:25] <mvita> ;-)
[13:56:39] <kaduk@jabber.openafs.org/barnowl> :)
[13:56:41] <meffie> oh, yes 13063 was the one i was thinking about
[13:56:45] <wiesand> heh
[13:59:55] meffie leaves the room
[13:59:56] mmeffie joins the room
[14:01:28] Marcio Barbosa joins the room
[14:02:10] <kaduk@jabber.openafs.org/barnowl> In terms of 1.8.1 tasks, someone(TM) still needs to go through the
libafsauthent/libafsrpc symbols and see if there are more to add to
the export lists.  Also actually making pullup changes in gerrit for
the stuff that's only on master...
[14:02:29] <wiesand> 13063 pullup to 1.8 is 13084
[14:02:58] <mmeffie> my memory continues to fail me! sorry.
[14:03:28] <kaduk@jabber.openafs.org/barnowl> I know how you feel.
[14:03:33] <mmeffie> kaduk@jabber.openafs.org/barnowl: how would we know which ones the mit python stuff needs?
[14:03:57] <kaduk@jabber.openafs.org/barnowl> "in Anders we trust"
[14:04:31] <mmeffie> ok, sounds good.
[14:04:33] <kaduk@jabber.openafs.org/barnowl> He already verified the bits that I merged this morning as good enough
for python-afs.  But there's no particular reason to think that's a
representative or complete use case.
[14:05:22] <mmeffie> i see. ok.
[14:07:37] <kaduk@jabber.openafs.org/barnowl> Backing up a bit, I didn't see any objection to "I wouldn't hold 1.8.1
for the volume release thing".  So it sounds like the plan continues
to be "1.8.1 as soon as the pieces can come into place"
[14:08:13] <mmeffie> ok
[14:10:03] <mvita> d'accord
[14:13:59] <wiesand> > Also actually making pullup changes in gerrit for
the stuff that's only on master...
[14:14:08] <wiesand> I'll see what I can do
[14:14:20] <kaduk@jabber.openafs.org/barnowl> thanks, wiesand; you're a lifesaver
[14:14:48] <mvita> Thank you.
[14:15:11] <wiesand> as long as it's trivial, your time shouldn't be wasted on that
[14:15:23] <mvita> <hoists a virtual frosty beverage in your honor>
[14:15:43] <wiesand> :)
[14:16:49] <wiesand> Looks like we're finished for today?
[14:17:11] <mvita> I have nothing further.
[14:17:29] <mvita> except "yay scan-build"
[14:17:41] <mvita> and "yay linux-rc builder"
[14:18:04] <kaduk@jabber.openafs.org/barnowl> indeed
[14:18:11] <wiesand> right - thanks for fixing it
[14:18:12] <mmeffie> wiesand, if you talk to pat, tell him thanks.
[14:18:19] <wiesand> ok
[14:18:26] <mvita> Little OpenAFS is finally growing up <sniff>
[14:18:45] <wiesand> er…?
[14:18:53] <mvita> nm, just being silly
[14:19:39] <wiesand> Let's adjourn then. Thanks a lot everyone!
[14:19:50] <mmeffie> have a good weekend
[14:19:54] <mvita> Thank you, have a good one
[14:20:07] mmeffie leaves the room
[14:20:18] wiesand leaves the room
[14:22:09] <kaduk@jabber.openafs.org/barnowl> thanks everyone!
[16:02:29] mvita leaves the room
[16:16:14] mvita joins the room
[18:20:39] mvita leaves the room
[18:30:05] mvita joins the room
[21:46:26] Marcio Barbosa leaves the room
[22:23:37] <mvita> for the vos release bug:  https://gerrit.openafs.org/#/c/13090/
[22:42:59] mvita leaves the room