- Friday, August 23, 2019

release-team@conference.openafs.org

Friday, August 23, 2019< ^ >

Room Configuration

Room Occupants

GMT+0
[12:03:38] mbarbosa joins the room
[12:30:55] meffie joins the room
[12:44:37] mvita leaves the room
[12:44:40] mbarbosa leaves the room
[12:44:50] mvita joins the room
[12:44:51] mbarbosa joins the room
[14:57:03] wiesand joins the room
[15:00:23] <meffie> good afternoon.
[15:01:07] <wiesand> Hello
[15:01:08] <kaduk@jabber.openafs.org/barnowl> greetings
[15:02:51] <wiesand> So, I finally uploaded 1.8.4pre1 yesterday
[15:03:26] <meffie> hooray!
[15:04:07] <wiesand> I think I still have to release the candidate volume, will do that in a minute
[15:04:29] <wiesand> Proposed web change in gerrit 13809
[15:04:31] <kaduk@jabber.openafs.org/barnowl> openafs 1.8.4~pre1-1 MIGRATED to testing today
[15:04:46] <kaduk@jabber.openafs.org/barnowl> candidate is replicated?
[15:04:53] <wiesand> nice timing then ;-)
[15:05:27] <wiesand> yes it is replicated
[15:06:00] <kaduk@jabber.openafs.org/barnowl> I wonder what I'm remembering not being replicated, then.
[15:06:41] <wiesand> I think you chose not to replicate the prerelease volumes mounted in candidate
[15:08:01] <kaduk@jabber.openafs.org/barnowl> Ah, that could be it
[15:09:22] <wiesand> Now it's visible in the RO path too. Please have a look at that and the web change proposal. I'd simply reuse the wording there in the announcement once agreed.
[15:09:54] yadayada joins the room
[15:10:19] <yadayada> Hello all, sorry for joining late
[15:10:39] <wiesand> np
[15:12:03] <meffie> ah, the 'remove dead code' commit hit master!! yay
[15:12:17] <kaduk@jabber.openafs.org/barnowl> I tweaked it a little, but yeah
[15:12:18] <wiesand> I tried to merge away Mike's web fixes yesterday too. Managed to get three done, then the first path conflicts struck…
[15:12:37] <kaduk@jabber.openafs.org/barnowl> I figured that given Andrew's disclaimer, it was getter to just do it
than to wait for a round-trip of discussion
[15:13:01] <meffie> wiesand: sorry, i can rebase.
[15:14:47] <wiesand> Thanks, but then we shouldn't leave those lingering for long again. But I guess we can agree on a "merge swiftly, fix fallout later" strategy again?
[15:16:20] <wiesand> In addition, we should fix the script(s) creating those pages to avoid re-adding these issues, or this exercise will be of limited use…
[15:16:50] <kaduk@jabber.openafs.org/barnowl> That would be clever, yes
[15:21:42] <wiesand> That's what I have on the stable series today.
[15:22:52] <kaduk@jabber.openafs.org/barnowl> I tried to catch up on a chunk of the master backlog, yesterday, and
made some progress.  Hopefully I can actually look at the newer stuff
this week, including the 1.9.0 NEWS that Mike updated last week, and
think about pushing the button for that.
[15:24:33] <kaduk@jabber.openafs.org/barnowl> I also left some review comments on things (mostly?) from Andrew, so
we'll be waiting another cycle to get agreement there.
[15:24:52] <meffie> wiesand: looks like you merged 13570 before 13569 and 13568? which would cause the merge conflict. i'll rebase.
[15:25:17] <kaduk@jabber.openafs.org/barnowl> So, let's see, all the linux 5.3 stuff got merged to master now.
[15:25:41] <meffie> yay! the builder for master was happy this morning
[15:26:07] <wiesand> "remove dead code" looks daunting - should this be pulled up (probably requiring quite a few changes and significant review), or will we live with more code skew in O(100) files?
[15:27:04] <kaduk@jabber.openafs.org/barnowl> Interestingly, the failure that debian had with openafs 1.8.2 and
kernel 5.2 seems to have been not fixed in 1.8.3, but only as part of
1.8.4pre1 -- the "correct include for ktime_get_coarse_real_ts64" that
we thought was just for documentation and clarity (was 13437 on
master)
[15:27:20] <kaduk@jabber.openafs.org/barnowl> well, it's dead code, so in theory nothing is going to need to touch
it.
[15:28:12] <kaduk@jabber.openafs.org/barnowl> Mike's buildbot mention reminds me of the windows builder updates --
Alejandro updated the local build script to actually return with an
error if the build failed ... so now all the builds are failing.
[15:28:41] <kaduk@jabber.openafs.org/barnowl> I don't remember if anyone has already submitted a fix for the error
yet, though
[15:28:48] <wiesand> mike: looks like I did - sorry for screwing that up
[15:29:27] <meffie> no problem! the rebase worked. going to push it to gerrit
[15:29:38] <kaduk@jabber.openafs.org/barnowl> I'm not sure I can actually recommend this, but since I fetch all
changes from gerrit, I can 'git log' a given change and see what its
parents are in order quite clearly
[15:30:05] <wiesand> fine, I'll just have to rebase 13809 then…
[15:32:10] <yadayada> Last week I mainly worked on aklog impersonate issue and shared couple of changes with Mike and Ben. After feedback from Mike and Ben will create gerrit reviews this week.
[15:32:28] <kaduk@jabber.openafs.org/barnowl> sounds good
[15:34:59] <meffie> openafs-web commits pushed to gerrit.
[15:35:43] <meffie> yadayada: sorry for the delay on the aklog questions.
[15:36:16] <kaduk@jabber.openafs.org/barnowl> I'm also sorry about the delay on the aklog feedback -- I was vey busy
this week with the IETF stuff
[15:36:46] <yadayada> np mike, I am almost done with my testing, once you confirm I will push the changes
[15:37:02] <wiesand> Ben: thanks for the comment on 13809 - had my linux blinders on again… should be "support for Linux on the PPC64 Little Endian platform" I guess
[15:37:14] <meffie> ben, i think Mark has a fix for the windows build error, or at least knows the breaking change.
[15:38:05] <kaduk@jabber.openafs.org/barnowl> I definitely say a link go by to the breaking change (from a security
release like a year or two ago, IIRC)
[15:38:10] <kaduk@jabber.openafs.org/barnowl> *saw
[15:38:20] <meffie> iirc, one of the NMakefiles needs an update for an include file.
[15:38:46] <meffie> yeah, the bu* changes
[15:45:41] <meffie> >  I'm not sure I can actually recommend this, but since I fetch all
changes from gerrit
can you share your .git/config gerrit stanza? (i had something at one point to do that :)
[15:46:05] <kaduk@jabber.openafs.org/barnowl> [remote "origin"]
        url = ssh://gerrit.openafs.org/openafs.git
        fetch = +refs/heads/*:refs/remotes/origin/*
        fetch = +refs/changes/*:refs/remotes/origin/changes/*
[15:46:25] <meffie> thanks
[15:46:31] <kaduk@jabber.openafs.org/barnowl> I think just the last line is the incremental addition for the changes
hierarcny
[15:46:59] <meffie> i did not know more than one fetch = could be stacked like that. nice.
[15:49:44] <yadayada> I have one query on RHEL8 support, is it right forum to discuss on that ? else will send a mail
[15:49:55] <kaduk@jabber.openafs.org/barnowl> sure
[15:51:00] <yadayada> I saw some mails from Jeff saying on RHEL 8 PAGs will not work. Mainly below is a snippet
[15:51:08] <yadayada> RHEL8 does include one major change that is going to require adjusting
the way that AFS tokens are acquired during login.  RHEL8 runs the GNONE
desktop as a user service launched by systemd.  If the user's home
directory is stored in /afs then GNOME must have access to the user's
AFS tokens.  However, since the GNOME process is launched from systemd
it will not have access to the user's AFS tokens if a PAG is created for
each user login session
[15:52:02] <yadayada> I need to try above test .. but have someone tried it on GNOME login PAG based tokens
[15:52:47] <yadayada> Is it we need to disable PAG creation when using PAM configuration ?
[15:54:13] <wiesand> My understanding is that it's either per-user tokens, not per-session (thus, no PAGs) or disabling that "systemd —user" stuff. The full implications of the latter are unclear to me.
[15:55:30] <kaduk@jabber.openafs.org/barnowl> Yeah, that's my understanding as well
[15:56:30] <yadayada> So we will have PAM authentication without PAG based tokens ?
[15:57:46] <kaduk@jabber.openafs.org/barnowl> We may need to collect more data; asking on -info might be the best
bet
[15:58:12] <wiesand> OTOH, it's claimed that any user can access any session keyring for his uid anyway (via ptrace). If that's true, PAGs have been a convience rather than a security feature on Linux for a long time.
[16:00:21] <yadayada> sure i think i will try some test on rhel 8 relating to it and will let you know my findings'
[16:00:38] <kaduk@jabber.openafs.org/barnowl> Thank you!
[16:00:59] <wiesand> Yes, that would be very helpful, thanks!
[16:01:15] <meffie> thanks yadayada
[16:01:15] <kaduk@jabber.openafs.org/barnowl> Switching to a different topic:
Is there a brief summary for the afs-dir-pkg changes?
[16:01:56] <wiesand> PS3 of 13809 should address Ben's remark and the issues addressed in the html-clean-up stack
[16:02:22] <meffie> afs-dir-pkg is bug fixes
[16:02:31] <meffie> too brief?
[16:03:08] <kaduk@jabber.openafs.org/barnowl> Nah, that's good enough for now
[16:03:30] <wiesand> too brief for me of course, but who cares ;-)
[16:03:57] <meffie> [194130.554591] kernel BUG at
/root/rpmbuild/BUILD/openafs-1.8.2/src/libafs/MODLOAD-3.10.0-862.14.4.el7.x86_64-SP/afs_dcache.c:1299!
[16:04:45] <meffie> i think those fixes were indented to fix panics in afs_dcache
[16:05:04] <meffie> intended
[16:05:26] <meffie> (from my notes, mark is out today)
[16:05:52] <kaduk@jabber.openafs.org/barnowl> Cool
[16:06:06] <kaduk@jabber.openafs.org/barnowl> I mean, not cool that we have panics, but fixing them is good
[16:06:20] <meffie> lol
[16:06:32] <kaduk@jabber.openafs.org/barnowl> I don't think I have anything else for today; are there other topics
people want to cover?
[16:06:54] <yadayada> nothing from my side
[16:07:03] <meffie> just to mention, andrew has started rxgk-phase-2 patch cleanup/refactoring
[16:07:22] <kaduk@jabber.openafs.org/barnowl> exciting!
[16:07:24] <meffie> he should be pushing patchsets to gerrit for review soon.
[16:08:02] <kaduk@jabber.openafs.org/barnowl> cool; I'll be on a lookout for those
[16:08:20] <wiesand> may I beg once more for review of 13809?
[16:08:32] <kaduk@jabber.openafs.org/barnowl> of course!
[16:09:36] <meffie> +1 LGTM
[16:09:59] <meffie> thank you wiesand
[16:10:13] <kaduk@jabber.openafs.org/barnowl> yes, thank you
[16:10:55] <yadayada> LGTM
[16:11:06] <kaduk@jabber.openafs.org/barnowl> motion to adjourn?
[16:11:38] <wiesand> thanks! I'll bring the web change live and send the announcement shortly then
[16:11:43] <kaduk@jabber.openafs.org/barnowl> yay!
[16:11:50] <meffie> fyi, i will be traveling and unable to attend next week, sorry.
[16:12:08] <kaduk@jabber.openafs.org/barnowl> okay, thanks for the heads-up
[16:12:13] <wiesand> thanks for the heads up
[16:12:17] <wiesand> motion sustained - thanks a lot everybody!
[16:12:22] <yadayada> Thanks
[16:12:29] <meffie> have a good weekend.
[16:12:30] <kaduk@jabber.openafs.org/barnowl> thanks everyone -- have a good weekend!
[16:13:36] wiesand leaves the room
[16:16:13] meffie leaves the room
[16:51:00] yadayada joins the room
[17:07:10] yadayada leaves the room
[17:59:53] yadayada joins the room
[18:08:36] mbarbosa leaves the room
[18:16:31] yadayada leaves the room
[18:37:28] mbarbosa joins the room
[19:21:37] yadayada joins the room
[19:22:03] mvita leaves the room
[19:22:03] mbarbosa leaves the room
[19:37:35] yadayada leaves the room
[19:53:46] yadayada leaves the room
[20:02:25] yadayada joins the room
[20:42:26] yadayada joins the room
[20:58:29] yadayada leaves the room
[21:14:57] yadayada leaves the room
[21:21:15] yadayada joins the room
[22:00:34] meffie joins the room
[22:01:29] yadayada joins the room
[22:16:47] yadayada leaves the room
[22:32:57] yadayada leaves the room
[22:35:41] meffie leaves the room
[22:41:23] yadayada joins the room
[23:22:00] yadayada joins the room
[23:37:49] yadayada leaves the room
[23:53:46] yadayada leaves the room