- Friday, March 5, 2021

release-team@conference.openafs.org

Friday, March 5, 2021< ^ >

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

Room Configuration

Room Occupants

GMT+0
[11:57:05] mbarbosa joins the room
[14:34:43] meffie joins the room
[14:52:58] Cheyenne joins the room
[17:00:44] <kaduk@jabber.openafs.org/barnowl> greetings
[17:01:20] <Cheyenne> Hello
[17:01:35] wiesand joins the room
[17:01:44] <wiesand> Hello
[17:02:20] <wiesand> The first meeting over my brand new fibre :)
[17:02:26] <kaduk@jabber.openafs.org/barnowl> !!!
[17:02:36] <kaduk@jabber.openafs.org/barnowl> Finally!
[17:02:36] <mvita2> congrats
[17:02:52] <meffie> woot. the electrons are so fast!
[17:02:59] <Cheyenne> photons
[17:02:59] <mvita2> um
[17:03:03] <mvita2> yeah
[17:03:14] <Cheyenne> photons make light work
[17:03:16] <meffie> oh they are electrons by the time the reach me.
[17:03:28] <meffie> wonder how that works?
[17:03:33] <mvita2> you aren't wrong
[17:03:45] <meffie> anyway, nice!\
[17:04:25] <wiesand> mike: it's called  photo diode
[17:04:44] <meffie> i know, i was making a bad joke. ignore me.
[17:05:34] <kaduk@jabber.openafs.org/barnowl> Yadav sends his regrets
[17:05:40] <meffie> ok, thanks.
[17:05:44] <wiesand> pity
[17:06:32] <wiesand> Not much news from my side today.
[17:06:54] <kaduk@jabber.openafs.org/barnowl> There is not much left for a pre1, though, right?
[17:07:17] <wiesand> I was about to update the NEWS change, but then couldn't resist putting that fibre to work…
[17:07:26] <wiesand> No, there's not much left.
[17:08:49] <wiesand> Update NEWS, get in Linux-next and Big Sur if feasible soon, merge the FBSD changes already pulled up and probably pull up a few more recently merged on master.
[17:10:01] <meffie> seems big sur changes are close.
[17:10:18] <kaduk@jabber.openafs.org/barnowl> I confess I haven't managed to internalize all of Andrew's latests
comments on 14431 yet.
[17:10:37] <kaduk@jabber.openafs.org/barnowl> But it seems like we do want another patchset to be uploaded?
[17:11:20] <meffie> i think so.
[17:12:11] <wiesand> The Linux 5.12 changes seems to be well on their way too.
[17:12:23] <Cheyenne> Waiting for more feedback on those.
[17:12:40] <kaduk@jabber.openafs.org/barnowl> Well ... I did put +2s on them
[17:12:53] <kaduk@jabber.openafs.org/barnowl> But I was planning to ask here if anyone else wanted to or should
review before they go in
[17:13:24] <Cheyenne> I would like to see any comments from Andrew
[17:13:39] <kaduk@jabber.openafs.org/barnowl> Okay, I'm happy to wait a bit to let Andrew comment.
[17:14:49] <wiesand> Andrew will not refrain from killing them even after I merged them on 1.8.x ;-)
[17:15:13] <wiesand> (which is good and appreciated)
[17:15:18] <Cheyenne> Your comment about obtaining the namespace at mount time I think would be the key in actually implementing a namespace aware openafs
[17:15:20] <kaduk@jabber.openafs.org/barnowl> indeed
[17:16:57] <kaduk@jabber.openafs.org/barnowl> Maybe I will remember to ask Wes about it if I see him in IRC
[17:18:33] <Cheyenne> Anyway -- just ran a master test build against latest pull from linux kernel repo (with the namespace patch), still getting a clean build.
[17:19:45] <wiesand> So, I'll try to drive 1.8.8pre1. Releasing it next Friday seems feasible and reasonable.
[17:19:50] <wiesand> Thanks Cheyenne.
[17:20:07] <kaduk@jabber.openafs.org/barnowl> Okay, so  looking at Stephan's list, linux-next is waiting on Andrew
to review, big sur is waiting on a new patchset, FBSD is "we get what
we get", and I don't think we talked more about NEWS.
[17:20:30] <Cheyenne> I'll pull in my patches into 1.8.x just to make sure 1.8.x builds cleanly
[17:20:47] <kaduk@jabber.openafs.org/barnowl> Thanks Cheyenne
[17:20:59] <wiesand> There are lots of valuable comments on NEWS to process and incorporate into a PS2. Expect me to do that R.S.N.
[17:21:14] <kaduk@jabber.openafs.org/barnowl> *nods*
[17:22:59] <kaduk@jabber.openafs.org/barnowl> Is that the cue for "on to master/1.9"?
[17:23:36] <wiesand> I think so.
[17:25:15] <kaduk@jabber.openafs.org/barnowl> I think mostly this week I pressed forward with FBSD stuff
[17:25:40] <kaduk@jabber.openafs.org/barnowl> But I did look at the linux-5.12 bits, that seem in good shape (and
reasonably simple as far as linux-compat patches go)
[17:26:34] <kaduk@jabber.openafs.org/barnowl> On one of the FBSD changes, I did have a question, on the off chance
anyone is familiar with afs_readdir_move()...
[17:27:12] <kaduk@jabber.openafs.org/barnowl> On some OSes, struct dirent includes a 'd_off' member, and I'm not
sure if the semantics of that are comparable across OSes
[17:27:29] <kaduk@jabber.openafs.org/barnowl> (But I honestly would be surprised if anyone just knew this off the
top of their head.)
[17:28:02] <Cheyenne> yes -- I think upstream Linux tried to make it a fairly mechanical change if the filesystem wasn't going to take advantage of the new parameters.  And looking at some of the changes where they do (like ext4) even that seems pretty straight foward.
[17:28:46] <kaduk@jabber.openafs.org/barnowl> Maybe Linux is getting more responsible about how they break things :)
[17:29:27] <meffie> sorry, i dont know afs_readdir_move. maybe mvita2 knows?
[17:29:40] <mvita2> no, I'm looking, but it's not familiar
[17:29:43] <Cheyenne> As long as you didn't have a swapfile sitting on a filesystem ...
[17:29:59] <kaduk@jabber.openafs.org/barnowl> No need to look right now re afs_readdir_move, then
[17:30:51] <kaduk@jabber.openafs.org/barnowl> (The short story about afs_readdir_move() is that it's conceptually
just a macro to help with the UIOMOVE to get the dirent bits into
userspace, though it's implemented as a function.)
[17:31:12] <kaduk@jabber.openafs.org/barnowl> Anyway, it looks like I also looked at the ih_PkgDefaults topic this
week, and had a locking question.
[17:31:54] <mvita2> ih_PkgDefaults:  i saw that and you are probably right but I haven't gotten back to that yet
[17:31:58] <kaduk@jabber.openafs.org/barnowl> Maybe I am too over-zealous about locking correctness, but I don't
really think that just a single unlocked global variable for "have we
initialized $thing yet" is actually safe in the general case
[17:32:16] <kaduk@jabber.openafs.org/barnowl> I think it was just last night, so I don't expect you to have done
anything with it :)
[17:33:03] <kaduk@jabber.openafs.org/barnowl> But I think that's all I did this week.
[17:33:33] <kaduk@jabber.openafs.org/barnowl> I expect to get approximately nothing done next week, since there's an
IETF (virtual) meeting, with sessions from 0400-1000h my time.
[17:33:47] <kaduk@jabber.openafs.org/barnowl> (Note that that overlaps with this meeting slot.)
[17:34:47] <kaduk@jabber.openafs.org/barnowl> Hopefully I would have time to review new versions of bigsur and/or
linux-5.12 patches, though.
[17:35:07] <meffie> excellent, thanks. good luck with IETF.
[17:35:29] <wiesand> OK, thanks for the head up. My pets for 1.8.8pre1 would be the linux-5.12 ones, others could wait for pre2 or final IMO.
[17:36:01] <kaduk@jabber.openafs.org/barnowl> thanks
[17:36:29] <Cheyenne> Thanks
[17:36:43] <wiesand> 14546 was pulled up this week. Buildbot's +1 took 27h, haven't found out yet why.
[17:37:02] <Cheyenne> There was an issue with some of the MacOS builders that stalled things
[17:37:04] <kaduk@jabber.openafs.org/barnowl> There was a pile of like 5 buildbot verifications that seemed to land
within a 20 minute interval; it was very unusual
[17:37:19] <kaduk@jabber.openafs.org/barnowl> since no single builder is fast enough to do that many builds in that
time
[17:38:48] <wiesand> We'll see what happens after the NEWS update.
[17:38:48] <kaduk@jabber.openafs.org/barnowl> Not directly related, but that does remind me that I should probably
get back to moving our services onto the new EL7 VM at MIT
[17:39:06] <meffie> yes, i have been meaning to ask about that.
[17:39:20] <meffie> seems the new server is down at the moment?
[17:39:32] <kaduk@jabber.openafs.org/barnowl> Hmm, I haven't actually checked recently :(
[17:40:26] <meffie> i wasnt able to reach it when i restarted the master to fix the hung macos builders.
[17:40:50] <meffie> (i have playbooks to configure both old and new servers)
[17:40:58] <kaduk@jabber.openafs.org/barnowl> That's not a good sign.  Did you just get a timeout or an active ICMP
unreachable?
[17:41:34] <meffie> honestly, i did not look closely, i just focused on the old one to get it backup.
[17:41:45] <kaduk@jabber.openafs.org/barnowl> understandable!
[17:42:09] <meffie> i'll look again and send mail.
[17:43:34] <kaduk@jabber.openafs.org/barnowl> I can get to the new machine at the moment, at least.  Could be a
firewall issue
[17:44:02] <wiesand> Administrivia: I think next week's meeting will be the last one before the US switch to DST.
[17:44:14] <meffie> we've been have a lot of issues with the SNA dns, it might have been that.
[17:44:49] <Cheyenne> Oh.. on the static-analysis commits, I have 5 that I'm looking at splitting up
[17:44:50] <kaduk@jabber.openafs.org/barnowl> Ugh, DST
[17:45:01] <kaduk@jabber.openafs.org/barnowl> Cheyenne: okay, thanks for the heads-up
[17:45:25] <kaduk@jabber.openafs.org/barnowl> Any other topics?
I am thinking maybe I'll cut 1.9.1 once linux-5.12 is in.
[17:45:38] <wiesand> Sounds like a plan
[17:46:03] <Cheyenne> (picked the 2 linux-5.12 commits into 1.8.x -- straightforward pick, got a clean build)
[17:46:21] <wiesand> I have no other topics for today. (Thanks again Cheyenne!)
[17:46:29] <Cheyenne> had one context conflict (no code changes)
[17:46:29] <wiesand> Adjourn?
[17:46:41] <kaduk@jabber.openafs.org/barnowl> motion seconded
[17:46:45] <Cheyenne> 3rd
[17:47:15] <wiesand> Let's adjourn then. Thanks a lot everybody!
[17:47:27] <kaduk@jabber.openafs.org/barnowl> Yes, thanks everyone!
[17:48:23] <meffie> have a good week all. stay safe.
[17:48:23] wiesand leaves the room
[17:48:58] meffie leaves the room
[21:38:32] mbarbosa leaves the room