- Friday, December 13, 2019

release-team@conference.openafs.org

Friday, December 13, 2019< ^ >

Room Configuration

Room Occupants

GMT+0
[12:00:02] mbarbosa joins the room
[14:43:30] meffie joins the room
[16:26:04] wiesand joins the room
[16:56:56] <meffie> good afternoon
[16:57:20] <kaduk@jabber.openafs.org/barnowl> greetings
[17:01:47] <kaduk@jabber.openafs.org/barnowl> I see a wiesand in the membership list today
[17:03:42] <meffie> greetings wiesand
[17:04:29] <wiesand> greetings
[17:04:52] <wiesand> my apologies for not showing up last week
[17:05:32] <wiesand> I actually was sitting at my desk, the running computer right in front of me
[17:05:56] <wiesand> Alas, with my chin on my chest, eyes closed…
[17:05:59] <meffie> no problem.
[17:06:21] <kaduk@jabber.openafs.org/barnowl> Agreed, not a problem.
[17:06:26] <wiesand> The same happened today - I just woke up 90m earlier ;-)
[17:07:04] <wiesand> I was good to see you work on current topics w/o me later :)
[17:08:35] <wiesand> I'm afraid I don't have a lot to talk about today - once more.
[17:09:17] <wiesand> The good news: I'll have two weeks off now, which Im going to spend at a very remote and quiet place.
[17:09:48] <kaduk@jabber.openafs.org/barnowl> I have to figure out what days to take as vacation from work -- my
wife is working so we're not travelling, but I need to get my
christmastime openafs sprint in :)
[17:09:49] <meffie> nice!
[17:10:01] <meffie> yay!
[17:10:22] <wiesand> But the network is supposed to be good, and that should allow me to catch up with openafs topics
[17:12:24] <wiesand> It would be good to know what your availability will be during the holiday season
[17:13:08] <kaduk@jabber.openafs.org/barnowl> I expect to generally be around.
[17:13:21] <wiesand> Great news
[17:16:00] <wiesand> I noticed a lot of activity lately. We seem to be homing in in Catalina in particular.
[17:16:26] <kaduk@jabber.openafs.org/barnowl> It's always a bad feeling to have a buildbot worker that's doomed to
failure because the OS-support patches haven't landed yet
[17:17:21] <meffie> heh
[17:17:30] <kaduk@jabber.openafs.org/barnowl> I also tried to get some work in on the ipv6-prep work (I guess that
was mostly for last week and not this week)
[17:17:40] <wiesand> Well it seems they actually needed some work.
[17:17:52] <kaduk@jabber.openafs.org/barnowl> Andrew left a couple notes on https://gerrit.openafs.org/#/c/13833/17
about potential further improvements, and I was a bit curious if
anyone had thoughts
[17:18:12] <kaduk@jabber.openafs.org/barnowl> some work, I guess, but not complet re-working
[17:21:55] <meffie> cheyenne and i have made good progress in the sockaddr conversion work. those patches are in github in our public repos, but it's work in progress.
[17:22:06] <kaduk@jabber.openafs.org/barnowl> That's exciting!
[17:22:19] <wiesand> that's about ipv6 prep?
[17:22:26] <kaduk@jabber.openafs.org/barnowl> yes
[17:22:42] <wiesand> so yes, exciting!
[17:22:47] <kaduk@jabber.openafs.org/barnowl> I also have on my list to look at the 'rxgk-phase2' stuff, the first
few of which look like they should be pretty straightforward
[17:22:55] <meffie> we have ported most places that use (host,port) to sockaddr (which is a union of sockaddr_in (4) and sockaddr_in6 (6))
[17:23:27] <meffie> focus is in the fileserver and cache manager, rx, and libauth for now
[17:23:55] <wiesand> sounds like solid progress - what's the main obstacle to overcome?
[17:24:40] <meffie> well, mainly, not breaking things :)
[17:25:01] <kaduk@jabber.openafs.org/barnowl> Main obstacle is likely maintaining mental focus through the drudge
work to not break things, yeah
[17:25:07] <wiesand> ah, what a surprise ;-)
[17:25:38] <meffie> the hope is to have a set of "boring" patches that that seem safe.
[17:25:58] <wiesand> what should be the target release for v6 support? 2.0? later? or maybe even 1.9?
[17:26:03] <kaduk@jabber.openafs.org/barnowl> I think for last time I had also taken a look at Mark's afs-dif-pkg
topic.  I don't think there was anything critically urgent in there,
so I'm not asking for him to reprioritize it, but would be nice to
know it's on his radar
[17:26:46] <kaduk@jabber.openafs.org/barnowl> Hmm, https://www.openafs.org/pages/no-more-des.html only reserves 2.0
for non-DES ciphers and says nothing about IPv6
[17:26:46] <meffie> the main issue that for someone new to the code base, it is far from obvious what to change, because some things will never be converted to ipv6 addresses (for wire compat, for example)
[17:27:34] <kaduk@jabber.openafs.org/barnowl> I think we probably should not be holding up anything else to wait for
v6 support, so either 1.9 or 2.0 or 2.1 is possible
[17:28:12] <meffie> yes, i think that's on Mark's radar. Mark has some TAP tests for dir.c too, which should be helpful.
[17:28:16] <kaduk@jabber.openafs.org/barnowl> We would want to lump vldb format changes for rxgk and v6 support
together, of course, though, and perhaps there is some risk that if
the rest of v6 support wasn't ready then we'd have something not quite
workable
[17:28:21] <kaduk@jabber.openafs.org/barnowl> Ooh, that would be nice, yes
[17:28:57] <meffie> for the TAP tests, he found a way to "inject" errors to test the dir.c code.
[17:29:21] <meffie> good stuff.
[17:29:33] <wiesand> I agree on not blocking 1.9 or 2.0 on IPv6. But that page is rather dated, and I thought 2.0 is supposed to be the first stable release with rxgk.
[17:29:45] <kaduk@jabber.openafs.org/barnowl> right
[17:30:55] <kaduk@jabber.openafs.org/barnowl> I think it was two weeks ago that I mentioned a couple of debian bugs
that had come in with build issues; neither turns out to be very
exciting.
[17:31:07] <wiesand> IPv6, if ready, and especially if it's unlikely to break V4, would fit in very nicely IMO. But again, we shouldn't block on it. I was just asking what we should strive for.
[17:31:39] <kaduk@jabber.openafs.org/barnowl> One was "you're not using the latest version" and the other was kind
of a debian-process thing, wanting to make sure that we use the same
compiler the kernel uses when we build kernel modules.  But the change
will probably not be localized to the openafs package
[17:31:53] <meffie> ah, ok, thanks.
[17:32:29] <wiesand> I noticed some fairly minor updates trickling in on Ubuntu 16.04 and 18.04 this week. Are these related?
[17:32:29] <meffie> that the same issue we had with the linux-rc buildbot.
[17:32:48] <meffie> (same gcc level as the one to build the kernel)
[17:33:55] <kaduk@jabber.openafs.org/barnowl> I'm not sure which ubuntu updates you're thinking about, actually
[17:34:34] <kaduk@jabber.openafs.org/barnowl> (I just updated my desktop from 16.04 to 18.04 and that would probably
mask any minor updates going on...)
[17:37:02] <wiesand> sorry I'll have to check - we're using the ppa packages here
[17:37:24] <kaduk@jabber.openafs.org/barnowl> Yeah, you kind of have to use the ppa packages
[17:40:45] <kaduk@jabber.openafs.org/barnowl> I'm probably about out of topics, other than noting that I still need
to put out a 1.9.1 dev release.  Anyone else?
[17:42:57] <meffie> andrew is working on an issue for the linux-native-mounts, so updates are coming soon (maybe today)
[17:43:18] <meffie> just an fyi for anyone testing the patches in gerrit.
[17:44:11] <meffie> also, is there anything i can do for 1.6.25 ?
[17:46:30] <wiesand> yes: wait for me to catch up :-(
[17:46:59] <meffie> will do.
[17:47:45] <meffie> i can rebase again if you want to drop some of the gerrits.
[17:48:14] <meffie> and i think i can help with NEWS if you like.
[17:49:19] <wiesand> There's a couple of changes I feel don't belong into the old stable series. I promised to mark those -2 a few weeks ago, but just to bring up the ones coming to mind right now: removing -settime and the ih_sync thread seems absolutely right, but not at this time in the 1.6 series.
[17:50:16] <meffie> ok, sounds good.
[17:51:52] <meffie> thanks all. that's all i had. please have a safe friday the 13th.
[17:51:59] <wiesand> I also still want to understand how we ended up needing non-trivial rebases to go forward on 1.6.x. It was not skew due to -settime or something. This will take an amount of time in one go I simply hadn't available in the past weeks.
[17:52:16] <wiesand> I'll drive on Sunday the 15th.
[17:53:12] <wiesand> Regarding your wish for a release before Xmas: I doubt we'll have a release. But a solid prerelease should be feasible.
[17:54:07] <meffie> i think the skew was due to we kept the sync thread in 1.6 for "compatibility" but it was removed before 1.8.0.
[17:54:37] <meffie> but yeah, you can see it from the histories.
[17:54:51] <wiesand> no, that doesn't explain why changes were pulled up and verified by buildbot but later couldn't be merged nor rebased
[17:55:14] <meffie> oh, ok.
[17:55:42] <wiesand> that frankly vexes me
[17:55:55] <kaduk@jabber.openafs.org/barnowl> If two changes that were originally (on 1.8) dependent on each other
were pulled up to 1.6 independently, then attempting to rebase the one
on top of the other for the 1.6 versions would be a non-trivial rebase
[17:57:49] <wiesand> right, but I think that still isn't the full truth
[17:57:59] <kaduk@jabber.openafs.org/barnowl> I can neither confirm nor deny
[17:58:15] <kaduk@jabber.openafs.org/barnowl> without putting in more effort to try to reproduce the various
cherry-picks/rebases
[17:58:16] <meffie> i guess i'm less surprised it could build but not be merged, that seems likely if it is not a fast-forward merge.
[17:58:37] <wiesand> like, maybe both had to be tweaked heavily to apply to 1.6 at all
[17:58:59] <wiesand> or, more likely, I goofed up
[17:59:26] <meffie> it was changed because 1.8.x is different than 1.6.x wrt to the sync thread.
[17:59:33] <wiesand> anyway, understanding what happened there frankly is my #1 priority regarding 1.6.x
[17:59:37] <meffie> ok
[17:59:50] <kaduk@jabber.openafs.org/barnowl> That sounds like something Mike could help with :)
[18:00:05] <meffie> maybe i'm making it worse :)
[18:00:35] <meffie> and i should look again, because i could be remembering things wrong.
[18:02:07] <wiesand> I should be able to find out. It will just take a couple of hours comparing changes etc. ;-)
[18:02:39] <meffie> that sounds fair. let me know if i can help, and i will try.
[18:02:56] <wiesand> In the end, a waste of time I agree. But I do want to know what went wrong, badly ;-)
[18:03:02] <wiesand> Mike: thanks
[18:03:47] <wiesand> Let's adjourn now. Thanks a lot!
[18:03:47] <meffie> you are welcome.
[18:03:57] <meffie> have a good weekend.
[18:04:05] meffie leaves the room
[18:04:12] <kaduk@jabber.openafs.org/barnowl> Thanks everyone, and have a safe friday the 13th!
[18:04:35] <wiesand> well it's almost over here :)
[18:04:38] wiesand leaves the room
[18:05:17] <kaduk@jabber.openafs.org/barnowl> That should make it easier, right? ;)
[18:49:04] mbarbosa leaves the room
[19:29:38] mbarbosa joins the room
[20:32:49] mbarbosa leaves the room
[21:02:14] mbarbosa joins the room
[22:04:39] mbarbosa leaves the room