Home
release-team@conference.openafs.org
Friday, February 26, 2021< ^ >
kaduk@jabber.openafs.org/barnowl has set the subject to: openafs release team
Room Configuration
Room Occupants

GMT+0
[11:55:03] mbarbosa joins the room
[12:33:05] mbarbosa leaves the room
[12:34:41] mbarbosa joins the room
[14:37:32] Cheyenne joins the room
[16:44:00] yadayada joins the room
[16:50:10] wiesand joins the room
[16:57:14] meffie joins the room
[16:57:23] <meffie> good day
[16:57:34] <kaduk@jabber.openafs.org/barnowl> Hi Mike
[16:58:52] <wiesand> Hello
[16:59:06] <kaduk@jabber.openafs.org/barnowl> Hmm, gerrit is feeling a bit slow
[16:59:12] <mvita2> oh, it's that time again already?
[16:59:20] <yadayada> Hello All
[16:59:28] <kaduk@jabber.openafs.org/barnowl> Mark: not quite yet
[16:59:38] <mvita2> <wait for it>
[17:00:50] <Cheyenne> Hello all
[17:01:04] <kaduk@jabber.openafs.org/barnowl> Mark: now it's time
[17:01:09] <kaduk@jabber.openafs.org/barnowl> Hi Cheyenne
[17:01:28] <wiesand> Ok, a first draft of the 1.8.8 release notes is out there, and I received lots of feedback (thanks!) but haven't been able to compile it into an update yet.
[17:02:50] <wiesand> There also was a lot of feedback on the not-yet-merged FBSD catch-up stack. Looks like we're waiting for some changes to be merged on master before it's truly useful.
[17:03:13] <kaduk@jabber.openafs.org/barnowl> Yeah, I think the pressure for that has accumulated enough to make it
hit the top of the list
[17:03:20] <Cheyenne> So .. a quick "news" item.. linux 5.12 broke master (and 1.8.x).  Though it's looking like a simple fix - new parameter in some of the inode_ops functions.
[17:03:39] <wiesand> The other loose 1.8.x ends:
[17:03:47] <wiesand> 1) 10831
[17:03:59] <wiesand> 2) Any news on Linux 5.12?
[17:04:11] <wiesand> 3) How's Big Sur doing?
[17:04:16] <kaduk@jabber.openafs.org/barnowl> [sorry to interject: I'm getting 502s from gerrit and plan to restart
once I can log in]
[17:04:25] <Cheyenne> Yes -- 5.12 breaks.  But might be an easy fix
[17:04:30] <wiesand> 4) Any other late additions wanted for 1.8.8pre1?
[17:04:55] <mvita2> oh my, gerrit is indeed sluggish
[17:05:12] <Cheyenne> (the hardest part is going to be the autoconf stuff I think)
[17:07:24] <kaduk@jabber.openafs.org/barnowl> (3) Big Sur is looking pretty good -- Marcio updated 14431 and the
changes look good.  I have maybe one comment open, and would like to
get Andrew's review as well.
[17:08:16] <kaduk@jabber.openafs.org/barnowl> Yay, gerrit is looking better post-restart
[17:08:32] <wiesand> Thanks for the Linux-next and Big Sur news (and work!)
[17:08:37] <mvita2> tx Ben
[17:09:18] <wiesand> better, yes, thanks Ben
[17:09:35] meffie leaves the room
[17:09:35] mmeffie joins the room
[17:09:54] <wiesand> re 10831 I'll have a look at the already existing 1.6.x cherry-pick - maybe that helps backporting it
[17:11:46] <wiesand> and I think that's what I have on 1.8 today - focus is on updating NEWS so that everyone is happy and then releasing 1.8.8pre1
[17:12:31] <kaduk@jabber.openafs.org/barnowl> I have not looked at the 1.8.8 NEWS (yet); should I?
[17:13:10] <wiesand> Of course ;-) But maybe wait a day or two for PS2
[17:13:25] <kaduk@jabber.openafs.org/barnowl> okay
[17:13:56] <wiesand> On to 1.9?
[17:14:47] <yadayada> relating to 14082, should we mention it avoids crashes also
[17:15:31] <kaduk@jabber.openafs.org/barnowl> Huh, interesting.
[17:17:21] <kaduk@jabber.openafs.org/barnowl> Um.  1.9.  I did some stuff; let me try to remember...
[17:18:12] <kaduk@jabber.openafs.org/barnowl> Cheyenne has been revitalizing some of the static-analysis stuff that
Pat Riehecky did a couple years ago (so long?  It doesn't feel like
it), and I've been trying to look at those
[17:18:59] <mmeffie> excellent!
[17:19:07] <kaduk@jabber.openafs.org/barnowl> I mentioned BigSur and the FBSD stuff earlier, but those are also
underway
[17:19:57] <mmeffie> yadayada's colleague gave a positive test report on the volscan perf fixes from marcio.
[17:20:09] <kaduk@jabber.openafs.org/barnowl> After merging 14519 I opened 14545 as a follow-up for something else I
noticed while reviewing
[17:20:19] <mmeffie> those would be nice for 1.9.1 (at least)
[17:20:38] <kaduk@jabber.openafs.org/barnowl> They also mentioned some local work they'd done for ignoring/excluding
some volumes, which sounds useful.
[17:21:06] <yadayada> volscan patches really helped in performance. Thanks. We would be creating gerrit for timestamp related filters
[17:21:09] <mmeffie> yes, those do sound useful. i dont know if they plan on providing patches.
[17:21:20] <mmeffie> excellent news!
[17:21:32] <kaduk@jabber.openafs.org/barnowl> yes, excellent news; thanks!
[17:22:01] <mmeffie> i was wondering if it makes sense to cache by default.
[17:22:25] <kaduk@jabber.openafs.org/barnowl> Past-me was kind of thinking we should, yes
[17:22:26] <yadayada> Only issue is memory footprint can go high
[17:22:27] <mmeffie> e.g. flip the flag from -cache-all to -no-cache
[17:22:48] <wiesand> 14082: I'd expect bad things to happen when cache I/O fails - will try to remember adding something about crashes though. Adding remarks like this one as a comment to 14540 is most helpful.
[17:22:48] <mmeffie> (or -disable-cache or some such)
[17:25:04] <kaduk@jabber.openafs.org/barnowl> I don't think I have much else to report on this week, though -- we
had back-to-back IESG telechats so my workload was pretty elevated
[17:25:42] <kaduk@jabber.openafs.org/barnowl> And next week is frantically trying to prepare for the IETF plenary
(virtual) meeting the following week.
[17:25:50] <yadayada> I have been in process of creating changes for signing kernel objects. will be done by next week
[17:26:13] <yadayada> for dkms I am trying to add some script based on https://gist.github.com/dop3j0e/2a9e2dddca982c4f679552fc1ebb18df
[17:28:48] <wiesand> Anything else to discuss today?
[17:29:24] <Cheyenne> Thanks for going through the static analysis stuff.
[17:30:34] <wiesand> IIRC one problem w/ Pat's changes was that they touched lots and lots of files each.
[17:30:51] <Cheyenne> Yes.
[17:31:36] <Cheyenne> If breaking any of them up would be helpful, let me know
[17:31:39] <kaduk@jabber.openafs.org/barnowl> And the static-analysis fixes do tend to need careful review
[17:31:49] <wiesand> Refactoring them into smaller chunks could very well help getting them merged. At the lates when it comes to pulling them up to stable.
[17:32:24] <kaduk@jabber.openafs.org/barnowl> Yeah, and if smaller chunks is better for stable we should probably do
them on master as well to keep things in sync
[17:33:54] <Cheyenne> I'll "sort" them by # of files changed and will try to split some of the larger ones into "part-1", "part-2" ... of some such
[17:34:14] <mmeffie> thanks
[17:34:16] <wiesand> Sure. It's a major objective not to make changes on stable deviate from their parents on master.
[17:34:58] <wiesand> Thanks Cheyenne.
[17:36:53] <kaduk@jabber.openafs.org/barnowl> yes, thanks
[17:37:34] <wiesand> To spread some good news (for me, that is), if things work out I shall receive my first covid vaccine shot monday morning :)
[17:37:54] <Cheyenne> +1
[17:37:58] <mmeffie> hooray!
[17:39:03] <mvita2> congrats
[17:39:06] <Cheyenne> So it's looking like the linux-5.12 fix is simply adding a new parameter to some of the inode op functions.  It's related to user namespace
[17:39:42] <kaduk@jabber.openafs.org/barnowl> Good news!
[17:39:45] <wiesand> Good news, thanks. Won't block 1.8.8pre1 on it.
[17:40:03] <kaduk@jabber.openafs.org/barnowl> (on both Stephan's vaccine news and the simplicity of linux-5.12
fixes)
[17:40:18] <Cheyenne> and it appears that you don't need to do anything with it (unless you want to).
[17:41:21] <Cheyenne> fs: make helpers idmap mount aware: https://lore.kernel.org/containers/20210121131959.646623-25-christian.brauner@ubuntu.com/
[17:42:53] <wiesand> I wonder whether that could help with per-container sysnames?
[17:44:23] <Cheyenne> Can't tell.  Looks like it's related to idmapped mounts and translating uid/gids
[17:45:36] <wiesand> Just a thought.
[17:45:41] <Cheyenne> Yeah.. does look like it might help with containers..
[17:46:26] <yadayada> looks interesting it will help with user namespaces, and uid/git mapping.  Will work well where root in container is not same as root on host and can handle filesystem permission.
[17:48:12] <wiesand> Looks like we're finished for today? Adjourn?
[17:48:28] <yadayada> nothing from my side
[17:48:33] <Cheyenne> Oh.. one thing that I've been working on is going through linux-kernel-func.m4 and trying to document when functions were added/removed.
[17:49:33] <mmeffie> thanks cheyenne.
[17:49:42] <mmeffie> motion to adjourn ?
[17:49:50] <mvita2> second
[17:49:53] <Cheyenne> 3rd
[17:50:07] <wiesand> Sustained. Thanks a lot everybody!
[17:50:12] <kaduk@jabber.openafs.org/barnowl> Thanks everyone!
[17:50:17] <yadayada> thanks all
[17:50:22] mmeffie leaves the room
[17:50:27] <Cheyenne> thanks.  Stay safe everyone.. have a good week
[17:50:37] wiesand leaves the room
[18:06:52] yadayada joins the room
[18:23:34] yadayada leaves the room
[19:03:56] yadayada leaves the room
[19:23:39] yadayada joins the room
[19:35:31] yadayada leaves the room
[22:05:00] mbarbosa leaves the room
[23:52:33] Cheyenne leaves the room