- Friday, January 19, 2018

release-team@conference.openafs.org

Friday, January 19, 2018< ^ >

Room Configuration

Room Occupants

GMT+0
[03:40:32] mvita leaves the room
[04:07:42] mvita joins the room
[04:56:42] mvita leaves the room
[13:57:59] meffie joins the room
[14:00:02] <kaduk@jabber.openafs.org/barnowl> good morning
[14:00:30] <meffie> good morning.
[14:01:44] wiesand joins the room
[14:01:54] mvita joins the room
[14:02:07] <wiesand> hi
[14:02:14] <mvita> greetings
[14:02:14] <meffie> good afternoon wiesand
[14:02:56] <wiesand> on countdown towards the weekend :)
[14:03:02] <mvita> yay
[14:03:10] <wiesand> and what a week it was
[14:03:50] <meffie> indeed
[14:03:54] <wiesand> meltdown/spectre would be bad enough
[14:04:20] <wiesand> but the way it's being handled by pretty much everyone just makes it perfect
[14:05:07] <wiesand> consequently, i had no time to spend on openafs
[14:05:23] <meffie> understandable
[14:06:01] <meffie> and sorry i was late on getting the notes out last week.
[14:06:07] <wiesand> but that's not so bad - I think we're mostly waiting whether Linux 4.15 will be released on Sunday, or there'll be an rc9
[14:06:17] <wiesand> thanks for writing them!
[14:06:27] <kaduk@jabber.openafs.org/barnowl> My time probably should be going towards fixing the client on debian
(old)oldstable after the meltdown/spectre kernel update broke things.
[14:07:53] <meffie> :(
[14:07:53] <wiesand> So regarding 1.6, I think the plan is to release 1.6.22.2 early next week if Linux 4.15 is released then
[14:07:59] <kaduk@jabber.openafs.org/barnowl> Though I guess it is reassuring about usage numbers that people climb
out of the woodwork to report the breakage
[14:08:30] <meffie> ah, good
[14:09:43] <mvita> "out of the woodwork", heh
[14:10:01] <wiesand> I'm not aware of any candidates for late inclusion in 1.6.22.2? Are there?
[14:10:21] <meffie> not for 1.6
[14:10:39] <meffie> that i know of.
[14:10:40] <kaduk@jabber.openafs.org/barnowl> I guess the freebsd stuff, if it gets pulled up appropriately.
[14:10:47] <kaduk@jabber.openafs.org/barnowl> But don't hold the release on my account.
[14:11:28] <kaduk@jabber.openafs.org/barnowl> (I guess I hadn't decided if I wanted to wait for reviews before
merging that on master, anyway.)
[14:11:41] <kaduk@jabber.openafs.org/barnowl> (12862/12863)
[14:13:08] <wiesand> there aren't too many ids left between i386_fbsd_120 and ia64_linux2
[14:14:00] <kaduk@jabber.openafs.org/barnowl> I think the story is that GCO allocates us chunks of 100 at a time
anyway ... not that the registry situation for these things is very
good at all.
[14:15:26] <mvita> oh?
[14:16:13] <kaduk@jabber.openafs.org/barnowl> Effectively openafs is acting as a free-for-all on sysname ID numbers,
but I think they are "supposed to be" allocated by the
grand.central.org registrar.
[14:16:49] <kaduk@jabber.openafs.org/barnowl> I have some memory of jhutz saying something about allocating chunks
of 100 at a time that makes such usage mostly okay, but it is not a
great situation to be in.
[14:17:01] <mvita> ah, understood
[14:17:11] <mvita> thanks.
[14:19:15] <wiesand> re 1.6.23, I saw that Andrew reviewed the ubik changes - that's a start
[14:19:22] <meffie> yay
[14:19:54] <meffie> mvita and i still need to review those
[14:20:01] <mvita> oh yeah.
[14:20:08] <kaduk@jabber.openafs.org/barnowl> I guess if we want the autoconf refactor in 1.6.23 someone needs to
prepare a pullup for 1.8.x first.
[14:20:34] <meffie> ok, great! i'll do that today.
[14:20:37] <wiesand> If you have a few minutes to kill, let me know what you think about openafs-stable-1_6_x (7784-v1) <https://gerrit.openafs.org/#/q/status:open+project:openafs+branch:openafs-stable-1_6_x+topic:7784-v1> vs. v1
[14:21:09] <mvita> That was a nice change, thank you Mike for doing it and Ben for merging it.  A nice New Year's gift.
[14:21:32] <kaduk@jabber.openafs.org/barnowl> Oh, I think I did look at the 7784 stuff
[14:21:33] <wiesand> If you have one minutre to kill, please look at 12867
[14:21:34] <mvita> I didn't see it a few days ago, so it was a surprise
[14:21:43] <mvita> looking
[14:23:24] <mvita> oh - 7784 is also a surprise.
[14:23:37] <mvita> a lot happened while I was on vacation
[14:23:48] <mvita> I'll try to not go on vacation any more, sorry.
[14:23:56] <mvita> ;-)
[14:24:27] <kaduk@jabber.openafs.org/barnowl> The 7784-v2 includes some generic cleanup that's not wrong to have,
but I don't expect us to be touching the rx event implementation on
1.6.x anymore, so I'm inclined to go with -v1
[14:24:31] <wiesand> it's the very point of vacations to miss a lot of stuff ;-)
[14:25:47] <mvita> 12867 +1
[14:25:52] <wiesand> thanks
[14:26:16] <wiesand> and yes, -v1 is simpler, so probably the better option
[14:27:46] <wiesand> anyway, it looks like the plan to have a reasonably early 1.6.23pre1 very soon after the 1.6.22.2 release could actually work
[14:27:46] <kaduk@jabber.openafs.org/barnowl> I guess I didn't +1 it at the time because I didn't take the time to
actually understand what the change was doing.
[14:29:56] <meffie> wiesand: andrew also reviewed 12247, which i feel is close to being ready for master, and should cleanly backport to 1.8 an 1.6
[14:30:36] <kaduk@jabber.openafs.org/barnowl> it's so much more work to only +1 12867 than +2 -- there's an easy
top-level button for +2 but I have to actually "reply" to just +1 ;)
[14:31:12] <kaduk@jabber.openafs.org/barnowl> I attempted to start looking at 12247 last night but was too tired to
do it justice.
[14:32:40] <meffie> yeah, it's not something to look at not well rested.
[14:33:42] <meffie> mvita sould review the revised 12247 as well
[14:33:56] <mvita> I will take a look, Mike.
[14:34:18] <meffie> yay
[14:34:34] <mvita> I was not well rested when I wrote it, so it all evens out
[14:34:48] <kaduk@jabber.openafs.org/barnowl> heh
[14:35:18] <mvita> quick side topic:
[14:35:39] <mvita> A small bit of good news from me: I found out
what was causing my emails to mailing lists
(release team, openafs-devel) to have
y
after my signature.
OS X has a setting ("smart quotes and dashes")
that changes a double-dash
into a hyphen ("emdash"). That was forcing my
double-dash signature separator to a hyphen.
This made my emails no longer encodable in
us-ascii and forced utf-8 with base64 encoding.
These factors apparently confuse the mailing list software
when it tries to find my sig in order to append
the mailing list footer. You'll note that
my "corrupted" emails never had a mailing list
footer.
I've disabled that OS X "feature", so my emails should be fixed now.
[14:37:02] <meffie> macos being "helpful"
[14:37:08] <mvita> yup.
[14:37:23] <mvita> APPLE PLEASE STOP HELPING ME.
[14:39:07] <meffie> anything else for 1.6.23pre1?
[14:40:46] <meffie> looks like i should backport 12667 ?
[14:40:50] <wiesand> There are still a few changes you pulled up in summer...
[14:41:30] <wiesand> 12667 12684..7
[14:41:47] <meffie> the ones in the last notes on openafs-devel are correct?
[14:42:04] <meffie> i mean the correct gerrit numbers.
[14:42:10] <wiesand> I'll have to check
[14:42:18] <meffie> ok, thanks.
[14:44:03] <wiesand> if it show up in https://gerrit.openafs.org/#/q/status:open+project:openafs+branch:openafs-stable-1_6_x  and has a # >= 12645, it's a candidate for 1.6.23pre1
[14:44:32] <meffie> ok, thanks
[14:45:36] <mvita> meffie: I remember the vldb timeout series, I'll review the pullups
[14:45:48] <mvita> 12667 doesn't seem familiar
[14:45:59] <mvita> but I can look at that too.
[14:46:11] <wiesand> Marcio bumped 12716 right after last week's meeting
[14:47:43] <mvita> <I have to drop out at the top of the hour>
[14:47:48] <meffie> i'll ask andrew if he has time to review 12716
[14:48:28] <wiesand> He said it's a fix for some issue unmasked by the other ubik changes…
[14:48:52] <meffie> understood
[14:49:01] <wiesand> If that's true, 1..6.23pre1 would block on it.
[14:49:58] <meffie> yes
[14:50:04] <wiesand> But that's all I have for today
[14:50:21] <meffie> ok, thank you wiesand
[14:51:32] <wiesand> Anything else to discuss today?
[14:51:37] <kaduk@jabber.openafs.org/barnowl> Not much on the 1.8 front -- we got one success report and one report
of issues building the srpm.  IIUC, Mike has made/pulled-up srpm fixes
that should take care of that issue.
[14:52:22] <mvita> there is also the gentoo panic reported against 1.8.0-pre4
[14:52:30] <meffie> a site building 1.8.0pre4 userspace rpms for testing on rhel 6 !
[14:52:32] <mvita> I have been looking at that on and off
[14:53:31] <kaduk@jabber.openafs.org/barnowl> So, staged for pre5 we've got those srpm build fixes, configure
refactor, Andrew's deadlock avoidance in check_Dentry_race (12868),
and I guess the ubik thing (12716)
[14:53:49] <kaduk@jabber.openafs.org/barnowl> It kind of sounded like the gentoo panic would also be manifestable in
1.6 as well, i.e., not a regression
[14:54:20] <kaduk@jabber.openafs.org/barnowl> Oh, and the freebsd sysname/OS support
[14:54:26] <mvita> well, that's a good point
[14:54:42] <mvita> but I don't have any 1.6x reports
[14:54:52] <mvita> and I have not tried to reproduce the problem
[14:55:16] <mvita> Since his kernel is 4.15-rc7, I suspected kernel_read stuff
[14:55:24] <kaduk@jabber.openafs.org/barnowl> I haven't tried to repro, either; need to fix debian first!
[14:55:31] <mvita> but I have mostly ruled that out now.
[14:55:49] <mvita> so I was going to ask for a config log next
[14:58:06] <kaduk@jabber.openafs.org/barnowl> Thanks for looking at it, even if it is off-and-on
[14:58:55] <mvita> I should probably open a ticket for it.
[14:59:08] <meffie> capital idea
[14:59:23] <mvita> okay, sorry, but I have to drop out now.
[14:59:43] mvita leaves the room
[15:00:10] <wiesand> Looks like we're done for today?
[15:00:32] <meffie> yes, thank you
[15:01:00] <wiesand> Let's adjourn then. Thanks a lot!
[15:01:09] <wiesand> Weekend,, sweet weekend …
[15:01:25] <meffie> have a good weekend
[15:01:25] wiesand leaves the room
[16:27:48] mvita joins the room
[17:11:18] meffie leaves the room
[18:19:34] meffie joins the room
[21:28:14] meffie leaves the room