- Wednesday, September 20, 2017

release-team@conference.openafs.org

Wednesday, September 20, 2017< ^ >

Room Configuration

Room Occupants

GMT+0
[12:34:38] meffie joins the room
[12:35:01] meffie leaves the room
[12:35:10] meffie joins the room
[14:51:31] wiesand joins the room
[15:00:01] <kadukoafs@gmail.com/barnowl772461F2> Hi everybody
[15:00:16] <mvita> Hi!
[15:00:19] <wiesand> Hello
[15:01:00] <wiesand> I'm still on vacation - my Internet connection is slow and flaky
[15:01:33] <wiesand> Obviously, I haven't pushed out 1.6.21.1 yet :-(
[15:01:57] <wiesand> One reason is that I didn't manage last week, and it's really tedious from here
[15:02:24] <wiesand> The other one is that I wonder whether we shouldn't include 12717
[15:03:00] <meffie> greetings
[15:03:12] <wiesand> Thoughts on the latter?
[15:03:42] <mvita> looking
[15:04:31] <mvita> yes, including that in 1.6.21.1 would be appropriate
[15:04:57] <kadukoafs@gmail.com/barnowl772461F2> +1 from here, yeah
[15:05:07] <meffie> yes, my +1 still stands
[15:05:09] <wiesand> OK, thanks. Let's include it then.
[15:05:29] <wiesand> With a little luck I should manage to get it out this week...
[15:06:34] wiesand joins the room
[15:06:56] <meffie> yay!
[15:07:06] <wiesand> (connection quality depends on the weather, time of day and some other factor I haven't discovered yet…)
[15:07:17] wiesand leaves the room
[15:07:38] <kadukoafs@gmail.com/barnowl772461F2> phase of moon?
[15:08:11] <wiesand> No, much shorter period
[15:08:51] <mvita> phase of the AC mains?
[15:10:08] <wiesand> No, longer period
[15:10:13] <kadukoafs@gmail.com/barnowl772461F2> :)
[15:10:17] <meffie> heh
[15:10:35] <mvita> ok, let's bisect
[15:10:38] <wiesand> Wind, maybe
[15:10:55] <meffie> anyway, before you ask, yes 4.14 will be interesting.
[15:11:20] <meffie> have some work in progress for that... may push to gerrit.
[15:11:21] <wiesand> Yes, saw the messages about new_sync{read,write}
[15:11:27] <mvita> and the linux builders are acting up again (failed shell_1)
[15:12:07] <mvita> stephan:  I haven't heard of that one yet - do you have any details?
[15:12:39] <wiesand> No, I just saw these in the builders' logs...
[15:12:43] <meffie> oops, sorry. i wonder if we are just losing messages, since the builds are done behind the buildbot slave/worker.
[15:13:26] <wiesand> Mike, are you talking about a different issue?
[15:13:50] <meffie> maybe. dont mind me.
[15:13:58] <kadukoafs@gmail.com/barnowl772461F2> I have long suspected that mail from the machine hosting
gerrit/buildbot-master somemtimes gets eaten along the way, but don't
know of a great way to track down exactly what's going on.
[15:14:58] <meffie> hmm.
[15:16:06] <meffie> at any rate, 4.14 work in progress.
[15:16:17] <wiesand> from the "failed shell_1" output from today:
[15:16:20] <wiesand> /home/buildbot/openafs/src/libafs/MODLOAD-4.14.0-999-generic-SP/osi_vnodeops.c:829:11: error: ‘new_sync_read’ undeclared here (not in a function)
   .read = new_sync_read,
           ^~~~~~~~~~~~~
/home/buildbot/openafs/src/libafs/MODLOAD-4.14.0-999-generic-SP/osi_vnodeops.c:830:12: error: ‘new_sync_write’ undeclared here (not in a function)
   .write = new_sync_write,
            ^~~~~~~~~~~~~~
scripts/Makefile.build:311: recipe for target '/home/buildbot/openafs/src/libafs/MODLOAD-4.14.0-999-generic-SP/osi_vnodeops.o' failed
[15:16:41] <mvita> OH.
[15:17:03] <mvita> I've been assuming that all "Failed shell_1" results were early timeouts - no actual build.
[15:17:08] <meffie> yes, we have a patch for that.
[15:17:11] <mvita> that was a very bad assumption.
[15:17:39] <mvita> now I see that we have in fact been building and having genuine build failures :-(
[15:17:58] <wiesand> Mike: Great. But you sound like there's more to it?
[15:18:28] <meffie> just havent had time to push it yet.
[15:19:53] <meffie> i dont think it will be too bad. the linux commit that broke our build is:
commit eb031849d52e61d24ba54e9d27553189ff328174
    Author: Christoph Hellwig <hch@lst.de>
    Date:   Fri Sep 1 17:39:23 2017 +0200
[15:22:34] <wiesand> The solutionis to switch to kernel_read/write?
[15:22:57] <kadukoafs@gmail.com/barnowl772461F2> presumably
[15:24:46] <wiesand> [hmmm, that mosquito protection lotion said "avoid contact with plastics or varnish" - but I didn't expect it to actually dissolve my Macbook…]
[15:24:55] <kadukoafs@gmail.com/barnowl772461F2> !!
[15:25:52] <wiesand> [but it works very well...]
[15:25:57] <meffie> oh, that stuff is toxic, and dissolves everything.
[15:26:52] <meffie> must be a good vacation if you need mosquito repellent!
[15:27:25] <wiesand> Well, it's scndinavia
[15:27:40] <wiesand> And I'm sitting outside...
[15:27:46] <kadukoafs@gmail.com/barnowl772461F2> so it must be good?
[15:28:29] <wiesand> It is good.
[15:28:39] <wiesand> But anyway, I think that's all on 1.6/Linux for now. On to 1.8?
[15:30:10] <mvita> Mike thanks for doing the Linux work
[15:30:38] <wiesand> +1
[15:30:44] <kadukoafs@gmail.com/barnowl772461F2> Yes, thanks for keeping on top of that
[15:31:23] <meffie> well, it's patches from our colleague in europe
[15:31:23] <kadukoafs@gmail.com/barnowl772461F2> I see that Mike +1'd 12723 for 1.8 ... when I merged it on master I
was not entirely sure that there was no behavior change in edge cases.
[15:32:11] <meffie> hmm, seemed save. i can look harder.
[15:32:11] <kadukoafs@gmail.com/barnowl772461F2> So it would be good to confirm that you're not worried about
breaking anyone or you think there's no change in behavior in the edge
cases.
[15:32:31] <meffie> * safe
[15:33:28] <kadukoafs@gmail.com/barnowl772461F2> I mean, you probably don't have to look too too hard.
[15:33:40] <kadukoafs@gmail.com/barnowl772461F2> Anyway, that's probably the less important issue for 1.8 that I know
of.
[15:34:33] <kadukoafs@gmail.com/barnowl772461F2> The other one is that closer inspection of the debian buildd test
failure seems to indicate an issue with rx event cancellation
[15:36:01] <kadukoafs@gmail.com/barnowl772461F2> Anders and I were eventually able to reproduce locally, by patching
the test to be less random about when to fire events:
diff --git a/tests/rx/event-t.c b/tests/rx/event-t.c
index 2812fa5bd..b9ba7c85c 100644
--- a/tests/rx/event-t.c
+++ b/tests/rx/event-t.c
@@ -127,7 +127,7 @@ main(void)
     /* Add 1000 random events to fire over the next 3 seconds */

     for (counter = 0; counter < NUMEVENTS; counter++) {
-        when = random() % 3000;
+        when = 1;
        clock_GetTime(&now);
        eventTime = now;
        clock_Addmsec(&eventTime, when);
I needed to use 1; he needed to set it to 0; so you may need to play
with it a bit in order to reproduce
[15:36:41] <meffie> oh, that test.
[15:37:14] <kadukoafs@gmail.com/barnowl772461F2> I didn't get to look closely at the core yet, but it seems that we try
to rxevent_put(NULL) because the cancellation zeroed out a structure
but we end up processing some part of it anyway
[15:37:25] <kadukoafs@gmail.com/barnowl772461F2> (very vague because my understanding is very vague)
[15:37:26] <wiesand> This issue is not present in 16.x?
[15:37:40] <kadukoafs@gmail.com/barnowl772461F2> 1.6.x has a different scheme for rx events, yeah
[15:37:48] <meffie> we dont have that test on 1.6.x if i recall.
[15:37:52] <kadukoafs@gmail.com/barnowl772461F2> Though it also doesn't have the test, yeah
[15:38:08] <mvita> 1.8 is rbtree, correct?
[15:38:48] <mvita> (uses rbtree to manage rx events)
[15:39:00] <kadukoafs@gmail.com/barnowl772461F2> Yeah
[15:39:42] <kadukoafs@gmail.com/barnowl772461F2> Anyway, I plan to look at it more (but have a freebsd thing to finish
first, so not in the next couple days at least), but it definitely
wouldn't hurt to have more eyes on it.
[15:40:43] <meffie> ok.
[15:40:55] <kadukoafs@gmail.com/barnowl772461F2> And that's all I have for 1.8 -- anyone hear of testing results from
any sites?
[15:40:57] <meffie> thanks for the test patch.
[15:41:39] <kadukoafs@gmail.com/barnowl772461F2> thank Anders :)
[15:42:03] <kadukoafs@gmail.com/barnowl772461F2> So if there's no other 1.8 topics, I guess we could move on to master.
[15:42:47] <meffie> still working on deploying 1.8 in a pilot...
[15:43:15] <kadukoafs@gmail.com/barnowl772461F2> Cool
[15:43:17] <meffie> no feedback yet
[15:43:29] <kadukoafs@gmail.com/barnowl772461F2> re master, I guess everyone's been pretty busy with other things, as
I don't remmber any comments on the rxgk patches.  (Some of them are
clean rebases from before and retain +1's from Chas, though.)
[15:43:53] <meffie> still on my radar.
[15:44:01] <kadukoafs@gmail.com/barnowl772461F2> *nods*
[15:44:16] <mvita> yes, sorry, extremely busy here
[15:44:23] <meffie> maybe andrew has some time to review. i'll ask.
[15:44:27] <kadukoafs@gmail.com/barnowl772461F2> along with dusting off the IPv6 stuff, too -- your todo list sounds
quite long, yeah
[15:45:45] <meffie> yes
[15:47:17] <kadukoafs@gmail.com/barnowl772461F2> Alright, well, we'll just have to see what we can fit in for this
week, then.
[15:47:19] <kadukoafs@gmail.com/barnowl772461F2> Any other topics?
[15:47:56] <meffie> not today.
[15:48:06] <wiesand> not here
[15:48:18] <mvita> plate spinning:  https://www.youtube.com/watch?v=Zhoos1oY404
[15:48:40] <mvita> nope, nothing here
[15:48:45] <mvita> ;-)
[15:48:53] <kadukoafs@gmail.com/barnowl772461F2> Thanks everyone -- let's adjourn
[15:48:59] <meffie> i should mention marcio fixed a bug in cellcc, the openafs-contrib program on github.
[15:49:01] <wiesand> Let's adjourn then.
[15:49:04] <wiesand> Thanks a lot everyone.
[15:49:12] <meffie> thanks!
[15:49:19] <wiesand> Got to wash off that repellent now...
[15:49:37] <meffie> have a good evening wiesand
[15:49:42] wiesand leaves the room
[15:55:27] meffie leaves the room
[17:07:33] meffie joins the room
[17:38:36] meffie leaves the room
[18:37:13] meffie joins the room
[20:01:27] meffie leaves the room
[20:40:11] meffie joins the room
[20:40:54] meffie leaves the room