- Friday, July 24, 2020

release-team@conference.openafs.org

Friday, July 24, 2020< ^ >

kaduk@jabber.openafs.org/barnowl has set the subject to: openafs release team

Room Configuration

Room Occupants

GMT+0
[00:06:26] meffie leaves the room
[00:07:39] meffie joins the room
[01:18:09] meffie leaves the room
[01:18:09] meffie joins the room
[02:29:33] meffie leaves the room
[02:29:33] meffie joins the room
[04:50:40] meffie leaves the room
[04:50:40] meffie joins the room
[04:51:10] meffie leaves the room
[04:51:11] meffie joins the room
[05:10:59] meffie leaves the room
[05:10:59] meffie joins the room
[05:11:28] meffie leaves the room
[05:11:29] meffie joins the room
[05:48:48] meffie leaves the room
[05:49:04] meffie joins the room
[06:09:56] meffie leaves the room
[06:09:56] meffie joins the room
[06:10:24] meffie leaves the room
[06:10:27] meffie joins the room
[06:31:13] meffie leaves the room
[06:31:13] meffie joins the room
[06:31:42] meffie leaves the room
[06:31:54] meffie joins the room
[07:36:57] meffie leaves the room
[07:36:57] meffie joins the room
[07:37:27] meffie leaves the room
[07:37:30] meffie joins the room
[11:16:24] meffie leaves the room
[11:16:24] meffie joins the room
[11:16:53] meffie leaves the room
[11:16:56] meffie joins the room
[11:35:12] meffie leaves the room
[11:35:12] meffie joins the room
[11:35:41] meffie leaves the room
[11:35:45] meffie joins the room
[11:37:18] meffie leaves the room
[11:37:18] meffie joins the room
[11:37:48] meffie leaves the room
[11:39:22] meffie joins the room
[11:47:30] meffie leaves the room
[11:47:31] meffie joins the room
[11:48:07] meffie leaves the room
[11:48:08] meffie joins the room
[12:05:24] mbarbosa joins the room
[13:21:33] mbarbosa leaves the room
[13:30:45] mbarbosa joins the room
[13:57:20] meffie leaves the room
[14:16:15] mbarbosa leaves the room
[14:16:16] mbarbosa joins the room
[15:29:02] meffie joins the room
[15:52:41] meffie leaves the room
[15:53:14] wiesand joins the room
[15:57:28] cwills joins the room
[15:57:56] yadayada joins the room
[16:00:06] <wiesand> Hello
[16:00:12] <cwills> Hello
[16:00:13] <kaduk@jabber.openafs.org/barnowl> Greetings
[16:00:17] <yadayada> Hello All
[16:01:07] mvita2 joins the room
[16:01:13] <mvita2> hello
[16:01:39] <wiesand> As usual, I'm a bit behind schedule. But I did manage to get something done this week, and it's more than probably meets the eye.
[16:01:42] <mvita2> Mike is on vacation and will not be joining us today
[16:01:50] <wiesand> Good for him!
[16:02:14] <kaduk@jabber.openafs.org/barnowl> Indeed, good for him!
My manager was just telling me this week that I should take more
vacation...
[16:02:36] <wiesand> Though this means I can't thank him for the notes and lists of proposed changes he sent.
[16:02:56] <mvita2> I will pass your thanks along to him
[16:02:56] <wiesand> Mark: Your manager may be right...
[16:03:36] <wiesand> So I went through these lists and what else was brought up in recent meetings:
[16:03:38] <mvita2> that was Ben's manager
[16:03:58] meffie joins the room
[16:04:13] <meffie> greetings from gary indiana
[16:04:15] <mvita2> meffie go home
[16:04:18] <wiesand> Ah, very sorry. Ben: Your manager is quite possibly right too ;-)
[16:04:36] <kaduk@jabber.openafs.org/barnowl> :)
[16:04:47] <kaduk@jabber.openafs.org/barnowl> Mike, you're on vacation
[16:04:50] <cwills> meffie go on vacation..
[16:05:06] <mvita2> not that we don't love having you here
[16:05:07] <meffie> carol is driving
[16:05:14] <meffie> heh
[16:05:33] <wiesand> When I'm on vacation, I can't actually relax unless I know things are going ok "at home".
[16:05:53] <wiesand> So welcome ;-)
[16:06:03] <cwills> tell me about it.. -- same here
[16:06:35] <wiesand> So I merged the override_creds changes. Thanks for the review.
[16:07:04] <meffie> thank you.
[16:08:14] <wiesand> I'm feeling uneasy about ubik-perf-18x and bulkstat-vcache-pressure. Both look like pure optimization to me, and I'm not convinced they're zero risk.
[16:08:21] <cwills> Hutzelman will be happy
[16:08:45] <wiesand> Well if he's not grumpy that's quite an achievement ;-)
[16:09:09] <cwills> s/happy/happier/    :)
[16:09:10] <meffie> ha!
[16:09:43] <wiesand> So I'm not saying I'll ban those stacks from 1.8.x, but they have lower priority for me than other changes.
[16:10:00] <meffie> should we submit any of the candidates to gerrit? i have the patch order.
[16:10:35] <wiesand> I actually looked through them and pulled up the low hanging fruit already.
[16:11:10] <meffie> thank you!
[16:11:31] <wiesand> That's those whcih look very reasonable to me, independent and straightforward to pull up. Reviewing those would be most helpful.
[16:12:02] <wiesand> Let me go through my list of changes I haven't pulled up yet and tell you why:
[16:12:36] <wiesand> 14066: path conflict with 14127
[16:13:09] <wiesand> and 14127, though clearly a fix, is a behaviour change in the middle of the stable series - do we want to do this?
[16:14:04] <wiesand> 14158: path conflict with 13718 (quite possibly trivial to resolve, haven't checked yet)
[16:14:23] <meffie> do we really care about "path" conflicts if the patches are in order?
[16:14:26] <kaduk@jabber.openafs.org/barnowl> Hmm, it doesn't look like I thought about whether or not to pull up
14127 when I merged it to master.
[16:15:00] <wiesand> 14153: no clean cherry-pick - not sure yet what's missing
[16:15:39] <wiesand> 14146/14147 fix ancient stuff like aklog -is this the time to do this?
[16:15:41] <kaduk@jabber.openafs.org/barnowl> Looking now, I think 14127 can be considered a bugfix that's
appropriate to apply in the stable series.
[16:16:02] <wiesand> 14181: looks like another mere optimization to me
[16:16:22] <meffie> well, i guess 14127 fixes the munged output. i'm not sure it's a compatibility issue though.
[16:16:25] <wiesand> 14203: will have to check/understand crosstalk with the aklog changes
[16:17:01] <wiesand> 14213: nice to have, but looks like low priority to me
[16:17:31] <wiesand> 14252: haven't gotten around to that yet
[16:17:32] <kaduk@jabber.openafs.org/barnowl> 14158/13718 are indeed trivial to resolve, and we actually want 14158
to fix problems.  13718 is maybe not so critical (though we might end
up getting it as part of the FBSD stack
[16:17:47] <mvita2> 14181 - at first glance looks exploitable as DOS to the fileserver?
[16:18:24] <wiesand> 13827: as said last week, there are many more aklog changes not yet on 1.8.x - will have to check for rxgk interplay for example
[16:18:45] <mvita2> ah, never mind
[16:19:03] <mvita2> I'm not caffeinated enough yet, apparenlty
[16:19:24] <kaduk@jabber.openafs.org/barnowl> I guess we should ask Yadav about 14146/14147: they are real bugs
getting fixed, but -setpag is perhaps not commonly used, and I don't
know if having the patches in official 1.8.x is useful for him.
[16:19:32] <wiesand> Haven't had a chance to look at FBSD catchup yet.
[16:20:17] <kaduk@jabber.openafs.org/barnowl> 14181: "optimization", while technically true, is perhaps misleading.
Without the patch, a misbehaving server can cause the whole client to
lock up for minutes.
[16:20:38] <wiesand> These are all about prioritization, not vetoes. Convince me that it's super-important and safe, and I'll shuffle priorities around.
[16:20:41] <mvita2> yes, that's it - I had it backwards at first
[16:21:12] <kaduk@jabber.openafs.org/barnowl> 14203 is completely orthogonal to aklog changes.
[16:21:30] <wiesand> Thanks
[16:22:02] <mvita2> re: 14181 - lossy network could also exacerbate the problem this fixes
[16:22:34] <yadayada> 14146 and 14147 are kind of regression in 1.8 compare to 1.6. So I think these changes should go in as 1.6 users will see this different behaviour
[16:22:47] <meffie> thank you for prioritizing, that is nice work!
[16:22:59] <kaduk@jabber.openafs.org/barnowl> Seconded, this is quite nice work.
[16:23:36] <kaduk@jabber.openafs.org/barnowl> and thank you for it!
[16:23:38] <wiesand> 14285 I did pull up, but again it's a behaviour change
[16:25:16] <wiesand> Points taken, thanks a lot for your feedback. It will drive my next steps.
[16:25:29] <meffie> i thought 14285 fixes a regression?
[16:25:46] <yadayada> 14285 , parallel option was not working and we always had 1 salvager process. and yes it looks as regression
[16:25:55] <meffie> ok thanks!
[16:26:20] <mvita2> i'm comfortable with behavior changes when the change brings actual behavior into line with documented / intended behavior
[16:26:29] <wiesand> If it is a regression against 1.6.x, fine.
[16:26:40] <meffie> yes, i think it is.
[16:27:08] <yadayada> 14146, 14147, 14285 all are regression compare to 1.6
[16:27:18] <wiesand> If we had this bug since 1.6.0, I'd rather change the documentation.
[16:27:45] <wiesand> yadav: got it, and that's important input
[16:29:04] <wiesand> That's about what I had on 1.8.7 for today. I'd still prefer to get pre1 out rather swiftly.
[16:29:25] <meffie> thank you wiesand
[16:30:05] <meffie> i'll add reviewers to the ones in gerrit (on the 1.8.x branch)
[16:31:19] <yadayada> Update from my side , I am still looking into NAT ping issue, had tried couple of change but those are not working in all scenarios. I will try couple of things next week
[16:31:44] <wiesand> Thanks
[16:32:14] <mvita2> I have a quick 1.6.x question if this is a good time
[16:32:32] <wiesand> There is no good time for 1.6.x questions ;-)
[16:33:05] <mvita2> does that mean you have no plans to make another release in 1.6.x?
[16:33:37] <mvita2> I'm asking re:  https://gerrit.openafs.org/#/c/13693/
[16:34:08] <mvita2> the problem it fixes has been observed at multiple sites but it is not a severe problem
[16:34:32] <wiesand> It means I'm painfully aware of how far behind 1.6.x is, and that it's my fault, but that I have no plans to commence work on 1.6.x until I've understood what went wrong back then.
[16:35:10] <mvita2> even if it never gets into a 1.6.x release, I wonder if could be merged for the benefit of those sites that need it.
[16:35:17] <mvita2> or want it
[16:35:28] <mvita2> they could pull it from gerrit
[16:36:29] <wiesand> I'm not going to merge anything onto 1.6.x before I understood the (sad) current state.
[16:36:41] <mvita2> okay, understood, thank you.
[16:37:06] <cwills> Kind of semi related.. any idea on updating the roadmap?
[16:37:41] <kaduk@jabber.openafs.org/barnowl> You refer to 14240?
[16:38:04] <wiesand> I'm open for a 1.6.x.1 (forgot: was it 23 or 24) branch starting at the last of my beloved security releases, and working on catching up from there.
[16:38:06] <cwills> yes
[16:38:38] <kaduk@jabber.openafs.org/barnowl> It looks like I was waiting for some nit-fixes on 14240 but would
otherwise be willing to merge it
[16:38:50] <wiesand> Alternatively, I already suggested reverting everything on 1.6.x back to the last security release and restarting from there.
[16:39:37] <kaduk@jabber.openafs.org/barnowl> Reverting back to the last security release (perhaps by a specially
crafted merge commit) is sounding like the most expedient option
[16:39:49] <kaduk@jabber.openafs.org/barnowl> But I forget exactly what the git history looks like at this point.
[16:39:58] <meffie> the sad current state of 1.6 is because the 1.6.x diverged from 1.8.x with respect to the ih_sync_thread
[16:40:21] <meffie> 1.6.x had patches that are not in 1.8.x on purpose
[16:40:46] <wiesand> I'd just revert them one by one. The resulting git history would still be completely linear, just as preferred by simple-minded folks like me…
[16:41:12] <meffie> you'd still have the ih_sync_thread divergence.
[16:41:17] <wiesand> Mike: No, that was not the culprit.
[16:41:27] <meffie> oh?
[16:41:50] <meffie> well, that was the issue i recall when i rebased last time.
[16:42:08] <meffie> but, i thing starting fresh is a great idea.
[16:42:14] <meffie> er think.
[16:42:53] <wiesand> The ih_sync_thread skew is real, and it will stay. But it can't possibly be the reason why at some point I was unable to merge or even rebase *anything* already pulled up to that branch.
[16:42:54] <meffie> in terms of what is on the openafs-stable-1_6_x branch and anything in gerrit for 1.6.x
[16:43:19] <meffie> hmm. yeah that sounds bad.
[16:45:01] <meffie> depending on the patch, significant "backporting" may be needed though, as the code bases are quiet different.
[16:45:19] <wiesand> That, and that I considered it more important to move 1.8.x along, is the reason why 1.6.x is that stuck.
[16:45:59] <meffie> i agree.
[16:46:07] <mvita2> agreed
[16:46:51] <meffie> i'd not put effort into 1.6.x at this point unless there is some very severe security issue
[16:47:26] <wiesand> In which case we'd probably branch off tha last security release anyway, right?
[16:47:33] <kaduk@jabber.openafs.org/barnowl> Yes
[16:47:40] <meffie> yes
[16:48:15] <meffie> is it crazy to just start a new branch at that point?
[16:48:31] <meffie> "1.6.x security fixes"  or some such?
[16:49:34] <wiesand> This is kind of what happened to 1.4.x, and it was a royal mess ;-)
[16:49:51] <meffie> ah, ok.
[16:50:40] <wiesand> (with the last coule of stable releases all coming from the 1.4.12.x branch - not pretty)
[16:51:07] meffie leaves the room
[16:51:24] <kaduk@jabber.openafs.org/barnowl> You get something kind of like the "new branch" thing with my
merge-commit idea
[16:52:01] meffie joins the room
[16:52:10] <wiesand> The problem is that I despise merges almost as much as security releases ;-)
[16:52:55] <kaduk@jabber.openafs.org/barnowl> Heh :)
[16:54:15] <wiesand> Anyway, enough on my woes for today - on to 1.9/master?
[16:55:04] <kaduk@jabber.openafs.org/barnowl> Nothing too exciting from here, I think.
A small flurry of things merged last night/this morning
[16:55:33] <cwills> still getting a clean build on linux 5.8 as of earlier this week
[16:55:42] <mvita2> thank you for the merges Ben
[16:55:42] <kaduk@jabber.openafs.org/barnowl> The firewall on the new machine at MIT got opened up some, but I
haven't had a chance to poke at apache yet
[16:55:53] <meffie> thanks bet
[16:55:56] <meffie> ben
[16:56:21] <kaduk@jabber.openafs.org/barnowl> You made it to the Chicago potholes, I see :)
[16:56:33] <meffie> heh
[16:56:39] <wiesand> Re the roadmap change: Did I mention that you should feel free to push over it?
[16:57:03] <kaduk@jabber.openafs.org/barnowl> You probably did, but thanks for the reminder.
[16:57:58] <kaduk@jabber.openafs.org/barnowl> I don't think there's anything else from me.  Just a reminder that
next week is (virtual) IETF 108 so I'll probably be a bit frazzled
[16:57:59] <cwills> I think having an updated roadmap will help some package managers as well as some users to know where things are going..
[16:58:04] <mvita2> fyi:  Marcio reports that he can no longer authenticate to gerrit.openafs.org via Yahoo OpenID;   likely due to changes on Yahoo's side:  https://developer.yahoo.com/oauth/social-directory-eol/
[16:58:18] <wiesand> :-/
[16:59:32] <cwills> I'm going to try to finish up the clang-10 stuff this afternoon (the f_fsid fixes)
[17:00:04] <meffie> good luck at ietf 108.
[17:00:16] <kaduk@jabber.openafs.org/barnowl> I'm happy to add new OpenID bindings to existing gerrit accounts via
the DB.  Just remember that it's many times easier to do so if you do
*not* try to log in with the new OpenID external account first, since
that creates a new gerrit account that has to be cleaned up
[17:00:46] <kaduk@jabber.openafs.org/barnowl> thanks re IETF 108!
[17:01:14] <kaduk@jabber.openafs.org/barnowl> IIRC github is an OpenID provider now, though I'm not sure if there
are any hoops to getting it to work with gerrit.  (Other than
gerrithub, that is.)
[17:01:48] <meffie> interesting.
[17:03:25] <kaduk@jabber.openafs.org/barnowl> Or maybe we need to enable the OAuth2 plugin on our gerrit side or
something.
[17:03:43] <kaduk@jabber.openafs.org/barnowl> Probably not something I'll think about until after we take a newer
gerrit, though.
[17:05:05] <mvita2> I've recently used github openid to authenticate to wireshark gerrit
[17:05:08] meffie leaves the room
[17:05:09] mmeffie joins the room
[17:05:22] <mvita2> I haven't tried it with our gerrit
[17:05:53] <mvita2> nor will I, now that I know it causes problems if I don't coordinate with Ben
[17:07:53] <mmeffie> motion to adjurn?
[17:08:03] <kaduk@jabber.openafs.org/barnowl> +1
[17:08:18] <mvita2> second
[17:08:19] <wiesand> Motion sustained. Thanks a lot everyone!
[17:08:20] <yadayada> +1
[17:08:22] <mmeffie> adjourn
[17:08:23] <yadayada> Thanks all
[17:08:31] <mmeffie> have a good weekend.
[17:08:41] mmeffie leaves the room
[17:08:54] <kaduk@jabber.openafs.org/barnowl> thanks everyone!
[17:09:31] wiesand leaves the room
[17:11:42] <cwills> thanks everyone .. have a good weekend
[17:11:47] cwills leaves the room
[17:14:52] mbarbosa leaves the room
[17:15:14] mbarbosa joins the room
[18:01:46] mbarbosa leaves the room
[18:01:48] mbarbosa joins the room
[18:13:29] mbarbosa leaves the room
[18:15:38] mbarbosa joins the room
[18:56:39] mbarbosa leaves the room
[19:01:57] mbarbosa joins the room
[19:38:28] yadayada leaves the room
[19:45:51] mbarbosa leaves the room
[19:45:55] mbarbosa joins the room
[21:43:36] mbarbosa leaves the room