Home
release-team@conference.openafs.org
Friday, August 14, 2020< ^ >
kaduk@jabber.openafs.org/barnowl has set the subject to: openafs release team
Room Configuration
Room Occupants

GMT+0
[12:18:00] mbarbosa joins the room
[12:21:25] meffie joins the room
[15:57:31] yadayada joins the room
[15:58:03] wiesand joins the room
[15:58:37] <meffie> good day
[15:58:44] kaduk@jabber.openafs.org/barnowl joins the room
[15:58:58] cwills joins the room
[15:59:17] <cwills> Good day everyone
[15:59:21] <yadayada> Hi All
[15:59:42] <wiesand> Hello
[15:59:50] <kaduk@jabber.openafs.org/barnowl> greetings
[16:01:40] <wiesand> I have largely been a slacker lately. I did finish giong through the latest pullups though.
[16:02:44] <wiesand> There are quite a few i'm inclined to merge after my next test build/run. Some could be reviewed by at least a second pair of eyes, though.
[16:03:52] <meffie> thanks wiesand
[16:04:06] <kaduk@jabber.openafs.org/barnowl> I'm happy that you are in a good state looking at pullups ... I'm in a
bad state looking at non-pullups
[16:05:18] <wiesand> I'm going to take refuge from the heat in the office over the weekend. Hopefully that will help get stuff done.
[16:06:18] <kaduk@jabber.openafs.org/barnowl> In my experience, having functioning air conditioning does wonders for
productivity in oppressive heat
[16:06:40] <kaduk@jabber.openafs.org/barnowl> (It's supposed to get to 35 or so here tomorrow)
[16:06:51] <meffie> heh. wiesand, andrew asked if we can consider gerrit 14255 for 1.8.7?
[16:07:25] <meffie> it fixes a situation where zeros are written to a volume when read() fails.
[16:07:35] <meffie> when restoring/release a volume.
[16:10:04] <wiesand> Hmm, that's a change in dump format.
[16:10:18] <meffie> hhm, i dont think so.
[16:10:41] <kaduk@jabber.openafs.org/barnowl> I had a similar reaction when we talked about it last week :)
[16:11:15] <kaduk@jabber.openafs.org/barnowl> I have since been mostly convinced that it is a change in
error-handling only, and that the previous behavior risks data loss
and is therefore buggy
[16:11:44] <meffie> yes, that's my understanding.
[16:11:54] <cwills> I believe that the commit addresses an issue when there is an error reading the dump file
[16:12:56] <meffie> on read errors, fail instead of silently dumping bogus data.
[16:12:57] <yadayada> last week I was not able to spend lot of time on Nat Ping. Recently we are seeing huge RSS value on dafileserver process and was looking at it.   Hopefully this week I will get time for NatPing.
[16:13:30] <meffie> with an option to override for manual recovery
[16:13:40] <kaduk@jabber.openafs.org/barnowl> "huge RSS value on dafileserver process" may be concerning in its own
right
[16:14:07] <wiesand> I put 14255 on the candidate list
[16:14:14] <meffie> thank you!
[16:15:28] <cwills> Just ran a build against the latest git pull on linux-5.9, still get a clean build with the 14300 fix
[16:15:39] <meffie> excellent
[16:16:10] <kaduk@jabber.openafs.org/barnowl> +1
[16:16:42] <cwills> Only comment about 14300 is the linux kernel "level set" to > 2.6.10
[16:17:51] <kaduk@jabber.openafs.org/barnowl> I don't have confidence that we work on 2.6.x less than 12-rc2 anyway
[16:19:09] <wiesand> That doesn't even affect RHEL5
[16:19:24] <cwills> RHEL4
[16:19:49] <kaduk@jabber.openafs.org/barnowl> "lalalalala I can't hear you"
[16:20:21] <meffie> ha!
[16:21:23] <cwills> Wait.. I want to run afs on my Mandrake system O_o
[16:21:39] <wiesand> While I don't like breaking compatibility with older platforms, ok with me.
[16:22:18] <cwills> Is there a spot where we could document the oldest supported kernel level?
[16:23:04] <kaduk@jabber.openafs.org/barnowl> The best thing I found last time I looked was NEWS, which is ... not
great for stable documentation.
[16:23:17] <kaduk@jabber.openafs.org/barnowl> Though I guess I didn't look at the website stuff
[16:25:03] <yadayada> Can't we have some check in configure step itself on kernel version and terminate configure step if kernel version < 2.6.10 ? just a thought
[16:26:33] <kaduk@jabber.openafs.org/barnowl> The name of the file src/cf/linux-guess-version.m4 suggests that might
not be a wholly reliable process
[16:26:55] <kaduk@jabber.openafs.org/barnowl> But at least issuing a warning if we think the kernel is too old is
still worthwhile
[16:27:11] <meffie> +1
[16:27:31] <kaduk@jabber.openafs.org/barnowl> Mike, is that you signing up to write the patch? ;)
[16:27:39] <meffie> heh!
[16:27:51] <meffie> sure!
[16:28:27] <meffie> i guess the lowest kernel level "supported" would depend on the branch, right?
[16:28:49] <cwills> (just checked .. 14300 clean build with clang-10 as well..
[16:29:06] <kaduk@jabber.openafs.org/barnowl> Technically, yes
[16:29:48] <meffie> maybe we could start with just a note in a README file, that gives a range of versions?
[16:30:01] <kaduk@jabber.openafs.org/barnowl> 1.6.x still claims to support linux 2.4, for example
[16:30:39] <kaduk@jabber.openafs.org/barnowl> range of versions as in oldest and newest?  I'm not sure that needing
to update for each newest version is something we'll remember to do
[16:31:16] <meffie> ah, good point
[16:31:22] <wiesand> which it probably doesn't
[16:31:40] <wiesand> how about INSTALL?
[16:32:05] <wiesand> Though I bet wherever we put it, it will rot quickly
[16:32:08] <cwills> We have the linux-rc builder on buildbot that kind of does that check.  Maybe having a builder that runs an older system might help..
[16:32:56] <yadayada> I do see some kernel version info in INSTALL as below
[16:33:04] <yadayada> ""
[16:33:05] <yadayada> The Linux 2.4 series (and older) are no longer supported.""
[16:33:21] <yadayada> So yes INSTALL could be good place to put version support info
[16:33:37] <meffie> yes, an update to INSTALL seems reasonable.
[16:34:14] <meffie> cwills, if you can create an ancient system, i'll add a buildbot worker for it.
[16:34:16] <wiesand> that should be part of 14300 IMHO
[16:34:37] <cwills> Okay.. say a RHEL5?
[16:34:47] <meffie> it doent need to change every night like linux-rc :)
[16:34:59] <meffie> er doesnt.
[16:35:02] <cwills> and I can tweek 14300 with an update to INSTALL
[16:35:39] <meffie> yes, I guess RHEL5 is the oldest version we can try to build and run on i suppose?
[16:35:49] <meffie> do others agree?
[16:36:13] <meffie> if it is fast enough, we can add it to the gerrit builders :)
[16:36:21] <yadayada> I agree with RHEL5
[16:36:22] <kaduk@jabber.openafs.org/barnowl> That seems like a plausible goal given our resources, yes
[16:37:12] <meffie> how does that sound wiesand?
[16:37:58] <wiesand> very good
[16:38:38] <cwills> Mike, I will work with you on the wording for the INSTALL update
[16:38:42] <meffie> great, we will have "bookends" for linux versions on the buildbot, RHEL5 to linux-rc (bleeding edge rc)
[16:41:19] <wiesand> More to discuss today?
[16:41:50] <wiesand> (I'll have to spend the remaining daylight on watering anyway)
[16:41:51] <kaduk@jabber.openafs.org/barnowl> I just have my usual "is there anything I should prioritize looking
at?" question.  Linux 5.9, I suppose.
[16:42:07] <wiesand> Right.
[16:42:46] <cwills> I'll update the 5.9 commit with an update the the INSTALL file about min version
[16:43:11] <meffie> thanks cwills.
[16:43:22] <wiesand> thanks!
[16:44:27] <kaduk@jabber.openafs.org/barnowl> So, motion to adjourn?
[16:44:42] <meffie> Ben, i was hoping for some love on 14154
[16:45:02] <meffie> and 14208
[16:45:06] <meffie> ?
[16:45:26] <kaduk@jabber.openafs.org/barnowl> I think those are still sitting open in browser tabs from two weeks
ago :(
[16:45:43] <kaduk@jabber.openafs.org/barnowl> But thanks for the reminder.  I think I will actually have some time
today (even though I didn't last night).
[16:45:55] <meffie> i've rebased (to fix the unrelated solaris buidlbot problem) and updated the commit messages, per andrew.
[16:46:07] <meffie> YAY!
[16:46:13] <kaduk@jabber.openafs.org/barnowl> thanks
[16:47:13] <wiesand> next motion to adjourn?
[16:47:30] <meffie> second
[16:47:39] <cwills> 3rd
[16:47:55] <yadayada> nothing from my side
[16:48:01] <wiesand> Fine then. Thanks a lot everybody!
[16:48:09] <yadayada> thanks
[16:48:25] <kaduk@jabber.openafs.org/barnowl> Thanks everyone; have a good weekend!
[16:48:32] wiesand leaves the room
[16:48:52] yadayada leaves the room
[17:08:09] cwills leaves the room
[17:42:33] meffie leaves the room
[18:02:30] mbarbosa leaves the room
[18:02:51] mbarbosa joins the room
[19:01:34] kaduk@jabber.openafs.org/barnowl leaves the room
[19:27:06] meffie joins the room
[20:32:02] meffie leaves the room
[21:42:56] mbarbosa leaves the room