- Friday, August 10, 2018

release-team@conference.openafs.org

Friday, August 10, 2018< ^ >

Room Configuration

Room Occupants

GMT+0
[00:14:49] meffie joins the room
[00:15:06] meffie leaves the room
[01:29:03] mvita joins the room
[02:16:56] mvita leaves the room
[12:11:24] Marcio Barbosa joins the room
[12:37:18] meffie joins the room
[12:52:20] wiesand joins the room
[12:55:25] <wiesand> good news - but he also commented that 1.6 is having problems… any news on that?
[12:58:28] <kaduk@jabber.openafs.org/barnowl> That's responding to Mark channeling Joe from last week post-meeting?
[12:58:39] <wiesand> er, yes
[13:00:10] <kaduk@jabber.openafs.org/barnowl> Okay, just checking I wasn't missing messages since then
[13:00:28] <meffie> good morning.
[13:03:04] <meffie> i'm not sure what joe is seeing.  it's not clear yet
[13:03:41] <meffie> or which versions he was testing.
[13:04:04] mvita joins the room
[13:04:15] <wiesand> Hello fols
[13:04:27] <wiesand> folks
[13:04:33] <mvita> present
[13:05:25] <wiesand> I merged everything foreseen for 1.6.23pre1 as discussed last week - anything missing?
[13:05:44] <wiesand> (except NEWS and version strings)
[13:05:51] <mvita> looking
[13:06:02] <wiesand> thanks for the reviews btw
[13:08:06] <mvita> that's weird - I gave a +1 to 11716 - I do NOT remember doing that!
[13:08:25] <kaduk@jabber.openafs.org/barnowl> "two years ago"
[13:08:45] <mvita> even though it happened ….. oh, yeah, NOT yeseterday
[13:08:48] <mvita> heh
[13:09:39] <wiesand> you cannot remember what you +1'ed in summer 16?
[13:09:52] <mvita> sorry to disappoint
[13:11:02] <wiesand> re this and 11706: I'm still reluctant to remove settime from the (now old) stable series
[13:11:20] <wiesand> I'm reluctant to un-break it either
[13:11:26] <mvita> yeah, it doesn't seem right
[13:11:46] <mvita> it's deprecated at the present time, correct?
[13:11:53] <mvita> (in 1.6.x)
[13:12:09] <wiesand> yes, and it has defaulted to off for years too
[13:12:19] <wiesand> and it's gone in 1.8
[13:12:20] <kaduk@jabber.openafs.org/barnowl> removing it entirely from oldstable seems like a bad plan, yes.
But it's less clear that fixing it is bad.  ("It's just a bugfix,
right?")
[13:12:43] <wiesand> it has been broken for all of 1.6 at least
[13:13:16] <mvita> 11706, plus whatever I would write for the other bug
[13:13:21] <wiesand> why fix it now then?
[13:13:33] <wiesand> so folks out there can start using it?
[13:14:23] <kaduk@jabber.openafs.org/barnowl> I don't think I can say this with a straight face, but "what about all
those people stuck on 1.4 because settime is broken in 1.6?"
[13:14:34] <mvita> HA
[13:14:41] <wiesand> RT #?
[13:15:22] <kaduk@jabber.openafs.org/barnowl> (I don't know of an RT entry, leading to the suspicion that "all those
people" comprises the empty set)
[13:15:41] <wiesand> seriously though, if you want to spend time on fixing it, go ahead :-)
[13:16:02] <meffie> 11706 is a bogus memset, regardless of afsd features
[13:16:17] <meffie> as far as i can see.
[13:16:24] <mvita> "Should one fix a null-pointer deref in kernel code for a deprecated option?  Discuss."
[13:16:46] <mvita> It is a crash of the OS for those who try it.
[13:16:52] <kaduk@jabber.openafs.org/barnowl> "But it's a root-only null-pointer deref, and root has plenty of ways
to crash the system already"
[13:16:57] <kaduk@jabber.openafs.org/barnowl> (not serious)
[13:17:06] <meffie> oh, you scared me :)
[13:17:26] <kaduk@jabber.openafs.org/barnowl> I mean, it's *true*, just not a serious argument
[13:17:34] <wiesand> It will show up in the changelog. And someone will start using it.
[13:17:42] <mvita> "Next time you'll believe it when something is deprecated"
[13:17:53] <kaduk@jabber.openafs.org/barnowl> It doesn't ... have to show up in the changelog.  Strictly speaking.
[13:18:01] <meffie> you dont have to mention the feature in the change log.
[13:18:28] <meffie> just that we fix setting a pointer to 0
[13:18:29] <wiesand> the changelog is autogenerated - it's just the git history
[13:18:58] <mvita> Last point - this code is so unexercised…. we've found 2 new bugs.  It is likely there are more
[13:19:13] <kaduk@jabber.openafs.org/barnowl> And we're above a little manual tweaking "[this change entry redacted
to prevent bad ideas]"?
[13:19:48] <wiesand> git log —oneline would still suggest we fixed settime
[13:19:53] <kaduk@jabber.openafs.org/barnowl> But if we go down Mark's route we should just make -settime a noop in
afsd and the syscall a stub?
[13:21:04] <mvita> ala 11716
[13:22:44] <mvita> let's do nothing.  It's already deprecated for good reason.
[13:22:51] <mvita> your gun, your bullet, your foot.
[13:22:58] <mvita> it's gone forever in 1.8
[13:23:34] <wiesand> doing nothing is what we've been doing for 3.5 years… no ill effects that I'd know of…
[13:24:04] <mvita> agreed, no reports
[13:25:06] <wiesand> ok, pre1 is pretty close now, finally
[13:25:20] <wiesand> I hope to get NEWS done over the weekend
[13:25:20] <kaduk@jabber.openafs.org/barnowl> tagging oracle stands ready
[13:25:34] <wiesand> thanks
[13:26:04] <wiesand> on to 1.8?
[13:26:20] <mvita> wait
[13:26:35] <wiesand> hmm?
[13:26:54] <mvita> should Ben -2 11716 and 11706 to reflect the conscensus?
[13:27:14] <mvita> apologies for spelling
[13:27:19] <kaduk@jabber.openafs.org/barnowl> (Can Stephan also -2?)
[13:27:25] <mvita> idk
[13:27:27] <kaduk@jabber.openafs.org/barnowl> Or would abandoning be more appropriate?
[13:27:39] <wiesand> I can -2 on 1.6.x
[13:27:55] <mvita> -2 is more accurate
[13:28:02] <wiesand> I'd leave them around as a reminder
[13:28:06] <mvita> yes
[13:28:16] <kaduk@jabber.openafs.org/barnowl> okay.
[13:28:17] <wiesand> like all other ones carrying a -2
[13:28:28] <kaduk@jabber.openafs.org/barnowl> sometimes reminders can get annoying, but that's definitely up to you
[13:28:53] <mvita> they can go away when 1.6 is sunsetted
[13:29:26] <meffie> mark -2 *and* abandon
[13:29:51] <mvita> Your call, Stephan
[13:30:00] <meffie> (gerrits can always be restored, so we should not be afraid to abandon imho)
[13:30:13] <meffie> (and then they dont show up in my status:open reports)
[13:30:16] <meffie> ;)
[13:30:46] <mvita> I just don't like the idea of unilateral abandonment - I think the author should do it.
[13:31:06] <mvita> Andrew wrote 11706
[13:31:14] <kaduk@jabber.openafs.org/barnowl> I try to only use my powers for good...
[13:31:32] <wiesand> I can't abandon changes I don't own.
[13:31:52] <meffie> ah, ok.
[13:32:13] <kaduk@jabber.openafs.org/barnowl> Well, if you want me to force-abandon things marked -2, you know where
to find me
[13:32:14] <wiesand> The release manager rules: We'll follow Mark's suggestion to do nothing :-)
[13:33:07] <meffie> heh.
[13:33:57] <wiesand> And unless they're rediscovered again (like this time by Marcio), they'll soon be in the "don't look at anything below) part of the changes display again.
[13:34:29] <wiesand> next try: on to 1.8?
[13:34:52] <mvita> +1
[13:35:34] <kaduk@jabber.openafs.org/barnowl> I merged the time64 thing for 4.18, but that's about it.
[13:35:55] <kaduk@jabber.openafs.org/barnowl> 1.8.1 final is in debian testing (as well as unstable); no bug
reports.  I didn't touch freebsd yet.
[13:36:39] <meffie> we got a report that we can now build 1.8.1 for rhel 5
[13:36:45] <meffie> rpms that is.
[13:36:55] <kaduk@jabber.openafs.org/barnowl> That's good to hear
[13:37:02] <wiesand> yay :-/
[13:37:15] <meffie> heh
[13:38:10] <kaduk@jabber.openafs.org/barnowl> OTOH, people should clearly be getting off RHEL5 posthaste.
But it's nice to know that the things we thought would fix an issue
worked as expected.
[13:38:13] <wiesand> will Joe be available to test things with 4.18 final?
[13:38:25] <meffie> yes
[13:38:52] <wiesand> great.
[13:39:20] <wiesand> and based on the results, we can then decide what to do next
[13:39:26] <kaduk@jabber.openafs.org/barnowl> So we think we will want a 1.8.1.1?
[13:39:50] <wiesand> (no point in shipping point releases which don't really work)
[13:40:26] <wiesand> but if they'll work, yes, I think we want 1.8.1.1 and 1.6.22.n+1
[13:40:30] <kaduk@jabber.openafs.org/barnowl> (sure)
[13:41:01] <wiesand> does that sound like a plan?
[13:41:04] <kaduk@jabber.openafs.org/barnowl> It's unclear whether we'll have enough to merit a 1.8.2 by then.  By
which I mean, looks unlikely from where I'm sitting now, but I could
be wrong.
[13:41:11] <kaduk@jabber.openafs.org/barnowl> So yes, that sounds like a plan.
[13:41:39] <kaduk@jabber.openafs.org/barnowl> I guess this is also an opportunity to ask you if you're in a place
where it would even make sense for me to ask if you are interested in
taking on the 1.8.x stable release manager mantle
[13:42:15] <wiesand> It's probably time to do that, yes.
[13:42:34] <kaduk@jabber.openafs.org/barnowl> Do you want me to send email or is here good enough?
[13:42:50] <wiesand> Here's good enough for me
[13:43:07] <kaduk@jabber.openafs.org/barnowl> Okay.  Should I make you a 1.8 release manager in gerrit?
[13:43:47] <wiesand> I'll have a few weeks off starting next Friday. I'll move home during that period, but there should be time left to get a grip on 1.8.x.
[13:44:15] <wiesand> Yes, please give me the superpowers.
[13:47:10] <mvita> Thank you Stephan!
[13:47:56] <wiesand> 1.8.x in it's current state will allow us to cut 1.8.1.1 w/o a branch. I'll leave it that way for the time being, until it's clear what we'll do about Linux 4.18.
[13:48:21] <wiesand> Thanks you for your support! I'll need it more than ever.
[13:48:56] <wiesand> On to master?
[13:49:06] <kaduk@jabber.openafs.org/barnowl> Yes, thank you Stephan!
(Still fighting with the gerrit permissions interface.)
[13:49:20] <wiesand> no hurry
[13:49:58] <kaduk@jabber.openafs.org/barnowl> master was pretty slow this week from my end; the TLS 1.3 spec keeps
dragging on and on.  It's like, "the RFC Editor is making what's
probably the final version", and that takes longer than I want, and
then we find things to change.
[13:50:32] <kaduk@jabber.openafs.org/barnowl> I did want to ask if there was a good merge order for the
solaris-warnings stuff; I merged several that look like the earliest
ones but wasn't sure if there were more ordering requirements.
[13:50:41] <kaduk@jabber.openafs.org/barnowl> Also I forget the story on the build failures :-/
[13:50:55] <wiesand> tls 1.3 is important to ~ everybody
[13:51:35] <meffie> ah, yes sorry about the merge ordering needed for the solaris-warnings. would it be ok if i rebased the remaining open gerrits?
[13:51:56] <kaduk@jabber.openafs.org/barnowl> Sure, please go ahead.
[13:52:02] <meffie> some ordering is needed.
[13:52:05] <meffie> ok, thanks.
[13:52:43] <mvita> Mike, thank you for that series - it will be so nice to have Solaris build warning-free
[13:52:45] <wiesand> there's also overlap (path conflicts) with solaris-kstat
[13:53:02] <meffie> it will be warning reduced.
[13:53:03] <kaduk@jabber.openafs.org/barnowl> IIRC the -warnings were higher priority
[13:53:21] <kaduk@jabber.openafs.org/barnowl> since there was the atomic change that fixes a bug on big-endian
arches
[13:53:30] <meffie> correct.
[13:53:51] <meffie> the kstats should be on top of the warning fixes.
[13:53:57] <mvita> yes
[13:54:06] <kaduk@jabber.openafs.org/barnowl> Feel free to include that in your rebase as well
[13:54:08] <mvita> kstats are much smaller and thus easier to rebase
[13:54:12] <wiesand> then kstats should probably be rebased on top of warnings
[13:54:22] <mvita> right
[13:54:54] <meffie> yes, i can rebase those too.
[13:55:09] <wiesand> the rebases will also yield a fresh set of buildbot results to look at
[13:55:17] <mvita> well, I just rebased them to master yesterday (kstats)
[13:55:22] <meffie> also, andrew has a solaris warning topic for vnop warnings
[13:56:16] <meffie> thanks, that will make it easier to keep the up to date then.
[13:56:20] <kaduk@jabber.openafs.org/barnowl> Hmm, gerrit is finding two of Stephan which is being problematic.
Maybe I'll have to come back to this later in the day (with direct sql
access).
[13:57:07] <mvita> hmm - cloning Stephan?
[13:57:20] <meffie> kaduk while you are doing gerrit admin, i have some buildbot / gerrit stuff i should ask about.
[13:57:24] <mvita> It always seems like a good idea at the time…
[13:57:25] <kaduk@jabber.openafs.org/barnowl> Anyway, I don't have much other news from master.
I *think* my tomorrow wil have several hours free, but we'll see.
[13:57:40] <meffie> i can put those questions in an email, minor stuff.
[13:57:52] <kaduk@jabber.openafs.org/barnowl> email or here is fine for me
[13:58:03] <kaduk@jabber.openafs.org/barnowl> but maybe we are wrapping up the meeting, for everyone else?
[13:58:27] <mvita> yes, another mtg at 10 for me and Mike
[13:58:46] <meffie> just a heads up here then;  we are using old ssh keys (dsa), and the email address the buildbot has uses jason's domain.
[13:58:49] <kaduk@jabber.openafs.org/barnowl> Ah, perfect timing.
[13:59:20] <kaduk@jabber.openafs.org/barnowl> Ah.  That email address thing does ring a bell, I think I even had an
idea for how to fix it :(
[14:00:05] <meffie> i just posted a notice to -devel about the buildbot transition. spent a lot of hours this week on porting our configuration to 1.3.
[14:00:19] <mvita> Thank you Mike
[14:00:19] <meffie> it's pretty close now.
[14:00:36] <wiesand> yes, thanks for all that work on buildbot
[14:00:52] <kaduk@jabber.openafs.org/barnowl> A huge thank-you to Mike for that!
[14:01:35] <meffie> it was cathartic
[14:02:44] <meffie> have a good weekend, notes shortly.
[14:03:06] <mvita> bye
[14:03:06] meffie leaves the room
[14:03:13] <kaduk@jabber.openafs.org/barnowl> Thanks everyone!
[14:03:26] <wiesand> Thanks a lot everybody! Have a nice weekend.
[14:03:31] wiesand leaves the room
[14:23:39] mvita leaves the room
[14:49:16] mvita joins the room
[18:11:26] mvita leaves the room
[18:42:07] mvita joins the room
[21:43:15] mvita leaves the room
[21:44:52] Marcio Barbosa leaves the room