- Wednesday, September 2, 2015

release-team@conference.openafs.org

Wednesday, September 2, 2015< ^ >

Room Configuration

Room Occupants

GMT+0
[13:26:08] meffie joins the room
[13:27:14] stephan.wiesand joins the room
[13:27:38] <stephan.wiesand> test test
[13:32:21] <Jeffrey Altman> ack akc
[13:51:29] stephan.wiesand leaves the room
[13:51:53] stephan.wiesand joins the room
[13:52:22] <stephan.wiesand> got kicked out
[13:56:51] kaduk joins the room
[14:00:55] <kaduk> $timeofday
[14:01:04] <stephan.wiesand> Hi
[14:01:41] <stephan.wiesand> Today's setup: http://www-zeuthen.desy.de/~wiesand/vmj.jpg <https://www-zeuthen.desy.de/~wiesand/vmj.jpg>
[14:02:23] <kaduk> "How can you manage to spend time looking at a computer screen instead of any of the other choices for things to look at?" ;)
[14:02:55] <stephan.wiesand> I just climbed a volcano. The break is quite welcome.
[14:02:57] <meffie> nice
[14:03:29] <stephan.wiesand> So I'm indeed on battery, and there isn't that much left.
[14:03:33] <kaduk> Volcano, sounds fun.
[14:04:12] <stephan.wiesand> Yes, the Eldfell erupted 1973. Grew the island by 20%. And buried 400 homes.
[14:04:42] <stephan.wiesand> Awesome view from the top. But quites a hike by my standrds.
[14:04:59] <stephan.wiesand> Ok. Linux news anyone?
[14:05:27] <stephan.wiesand> Are we still good for the final 4.2?
[14:05:27] <kaduk> I hear 4.2 released.
[14:06:08] <stephan.wiesand> That's right. Now if we knew that there weren't any last minute changes to the kernel, we could release 1.6.14.1.
[14:07:23] <stephan.wiesand> I'll ask Marc by PM then.
[14:07:34] <stephan.wiesand> Any comments on the NEWS change?
[14:08:23] <stephan.wiesand> (11993)
[14:08:26] <kaduk> Hmm, it doesn't quite convey the crux of the issue.
[14:09:20] <stephan.wiesand> It was the best I could come up with yesterday evening. Help in doing it better is very welcome.
[14:09:27] mvita joins the room
[14:09:39] <mvita> sorry I'm late...
[14:09:52] <stephan.wiesand> ts ts ;-)
[14:09:59] <kaduk> namely, that they must behave like symlinks in order to adhere to the linux VFS invariant about only one path to a given dentry, which is incompatible with AFS's scheme for having arbitrary mountpoints (assuming the current worldview where AFS mountpoints don't show up as linux mountpoints)
[14:10:13] <kaduk> (that's probably too much detail for the NEWS entry, and maybe not even 100% correct)
[14:10:27] <stephan.wiesand> yeah ;-)
[14:12:15] <stephan.wiesand> I'll try to rephrase it later. Maybe Marc can help too.
[14:14:04] <stephan.wiesand> 11989 needs review. Help appreciated...
[14:14:45] <mvita> looking
[14:15:14] <kaduk> "Looks pretty similar to the master change"
[14:15:36] <stephan.wiesand> well it ought to...
[14:16:18] <stephan.wiesand> But that doesn't guarantee it's right for 1.6...
[14:17:50] <kaduk> I don't expect anything interesting to have changed in its surroundings.
[14:18:24] <stephan.wiesand> That's what I wanted to hear :)
[14:19:02] <mvita> I'll review it after we finish here.
[14:19:11] <stephan.wiesand> I'll run smoke tests on EL5/6/7 as usual before releasing 1.6.14.1.
[14:19:16] <stephan.wiesand> Mark: Thanks.
[14:19:49] <mvita> refresh my memory - I remember you've talked about smoke tests before but what do they consist of?
[14:20:06] <stephan.wiesand> So, together with 11982 that will hopefully be 1.6.14.1.
[14:20:36] <stephan.wiesand> Mark: client builds, loads, and survives some basic use.
[14:20:51] <stephan.wiesand> "no obvious breakage"
[14:21:54] <stephan.wiesand> I know our testing is less than ideal.
[14:22:24] <stephan.wiesand> Our releases get actually tested when people start using them.
[14:23:15] <Jeffrey Altman> according to Marc there were few fixes to 4.2 after rc8 and they do not involve the vfs layer.   Marc has not explicitly tested against 4.2 final but he expects there to be no change in behavior
[14:23:24] <kaduk> The diffstat between v4.2-rc6 and v4.2 doesn't look to have any VFS changes in it that we care about.  I guess it's always possible that there was an allocator change or something, but it seems unlikely.
[14:23:47] <stephan.wiesand> Ok thanks.
[14:25:05] <stephan.wiesand> Ben: Any FBSD build fixes you want in 1.6.14.1?
[14:25:26] <kaduk> Uh, I think the only thing ready is that stray underscore, which may not be relevant.
[14:25:52] mvita leaves the room
[14:25:53] <kaduk> Yeah, it's not relevant.
[14:26:06] <kaduk> So, no FBSD fixes needed in 1.6.14.1
[14:26:13] <stephan.wiesand> Ok, then it can wait. Fine.
[14:26:28] <stephan.wiesand> On to 1.6.15 then: what must go in?
[14:28:06] <stephan.wiesand> I guess we'll want to release 1.6.15 about in time for Linux 4.3. Giving us a few weeks before we need a pre1 ready.
[14:28:07] <meffie> the shake-loose-vcaches bug continues to bite people.
[14:28:23] <meffie> yadav says he hit this bug as well.
[14:28:34] <kaduk> I'd like to see, uh, 11588 and 11715 with my debian hat
[14:28:45] <kaduk> 11894 seems like a good candidate
[14:30:18] mvita joins the room
[14:30:36] <stephan.wiesand> Yes 11894 is on my personal list ;-)
[14:30:52] <mvita> sorry, network woes
[14:30:53] <kaduk> We might as well do the volume-clone bits, since we went to all the effort of getting them (we think) right.
[14:31:18] <stephan.wiesand> shake-loose-vcaches-bug = the client consumes lots of slabs w/o freeing them as it's supposed to?
[14:31:48] <stephan.wiesand> Those from Hans-Werner?
[14:31:58] <meffie> yes, then when the log jam breaks, the machine hangs on the xvcache lock.
[14:32:46] <stephan.wiesand> volume-clone: yes
[14:34:00] <stephan.wiesand> how about 11594?
[14:34:11] <Jeffrey Altman> There are likely to be substantial changes in the Linux 4.3 vfs layer
[14:34:21] <kaduk> I am not seeing it in gerrit, but I kind of want to add a -help option to afsd, or at least something that I can pass to get it to exit success and do nothing, in order to make my debian systemd unit file less sad.
[14:34:34] <stephan.wiesand> Jeff: thanks for the "great" news.
[14:35:03] <kaduk> 11594 is probably reasonable, but I'd have to look at it again.
[14:36:18] <stephan.wiesand> That's probably enough fodder for a few weeks time.
[14:36:45] <stephan.wiesand> No objections ot a -help (-dryrun?) switch or something
[14:36:59] <Jeffrey Altman> I'm fine with -help
[14:37:08] <mvita> afsd does have -help but it's broken on master
[14:37:18] mvita leaves the room
[14:37:20] <kaduk> (I currently have patched debian to use -exit, which is not really appropriate for upstream)
[14:37:25] mvita joins the room
[14:38:06] <stephan.wiesand> -help is broken on master only?
[14:40:44] <stephan.wiesand> Ok, my battery is running low. Anything else to discuss today for which I should be present?
[14:40:47] <mvita> yeah, a recent change to the parser
[14:41:29] <stephan.wiesand> If it works on 1.6, why does Ben need a patch?
[14:41:57] <kaduk> I forget why it was unsuitable; maybe it didn't exit 0.  I'll try to remember to test and figure out what I actually want.
[14:43:02] <stephan.wiesand> Sounds good.
[14:43:27] <stephan.wiesand> Battery is really low now, and I have another volcano to climb...
[14:43:41] <kaduk> Have fun and take some pictures to taunt us with next week!
[14:44:00] <stephan.wiesand> I'll work on 1.6.14.1 in the evening.
[14:44:13] <stephan.wiesand> Ben: sure.
[14:44:27] <stephan.wiesand> Thanks a lot for participating today!
[14:44:39] <stephan.wiesand> Goodbye.
[14:44:43] stephan.wiesand leaves the room
[14:45:59] <kaduk> On the 1.8 front, I was skimming through the externalize-log-rotation changes in gerrit (just tracing the dependencies chain back), and it looked like there were only a couple that had negative review, and there's even some changes at the early part of the chain that could potentially be merged.
[14:46:23] <kaduk> I don't know what if any plans Mike has regarding updates for the review comments.
[14:46:26] <meffie> excellent. i have a new patch set for "convert to opr softsig"
[14:46:49] mvita leaves the room
[14:47:02] <kaduk> But since it's such a long chain I'm worried that I'll mess something up if I try to do a rebase and/or tweak anything myself.
[14:47:10] <meffie> i just need to finish the commit messages and then i'll push it to gerrit.
[14:47:22] <kaduk> Yay!
[14:47:54] <kaduk> I also wonder if I can ask folks to look at a couple of the -1'd changes and we can see if there are things to talk about here.
[14:48:07] <kaduk> Say, 11818.
[14:48:16] <meffie> then i can fix the -1's on the externalize-log-rotation topic
[14:48:53] <kaduk> IIRC, I created that because I made some review comments on some other change, but I didn't think through it very much at the time and am not sure whether it actually makes sense.
[14:49:48] <kaduk> (There's also 11390, which as far as I can tell is just waiting for Chas's people to come up with license text.  Though, we haven't heard much from Chas recently at all...)
[14:50:13] <meffie> i dont remember 11818. i'll star it.
[14:51:21] <kaduk> Okay.  Jeff or Daria, do you have any thoughts about 11818?
[14:53:59] <kaduk> I guess that's a "not right now"...
[14:55:13] <meffie> i'll ask mark v to look too. i think he's done some work in this area too.
[14:55:19] <kaduk> And since Mike already has a number of things on his plate "for me", I guess I'll stop here for this week and not add more things to review.
[14:55:23] <kaduk> thanks!
[14:55:38] <meffie> thanks ben
[14:56:18] <meffie> have a good afternoon.
[14:56:34] meffie leaves the room
[14:57:38] kaduk leaves the room
[14:58:40] <Jeffrey Altman> 11390 is unlikely to get a copyright update.   Chas resigned from NRL quite a while ago and no longer works with AFS
[15:00:45] <Jeffrey Altman> I'm looking at 11818
[15:05:12] <Jeffrey Altman> indirectly related to 11818, shouldn't SetSalFlag() be checking for a failure of "open()"?
[15:06:14] <Jeffrey Altman> at the moment the only error returned is from asprintf().   If open() fails, the file will not be created and SetSalFlag() returns success
[15:09:12] mvita joins the room
[15:12:24] <mvita> sorry for the flaky network, I was working offsite this morning.
[16:00:48] sorry, btw. just finished catching up. i was at therapy
[16:01:12] oh. sigh.
[18:26:39] mvita leaves the room
[18:26:39] mvita joins the room