- Wednesday, October 8, 2014

release-team@conference.openafs.org

Wednesday, October 8, 2014< ^ >

Room Configuration

Room Occupants

GMT+0
[05:03:29] Jeffrey Altman leaves the room
[05:11:46] Jeffrey Altman joins the room
[05:30:37] Jeffrey Altman leaves the room
[05:30:43] Jeffrey Altman joins the room
[06:09:04] Jeffrey Altman leaves the room
[06:10:55] Jeffrey Altman joins the room
[11:06:05] wiesand joins the room
[13:21:58] shadow@gmail.com/barnowlE5B64A04 joins the room
[13:34:27] shadow@gmail.com/barnowlE5B64A04 leaves the room
[13:38:54] <Jeffrey Altman> I will be unable to attend todays meeting.
[13:44:55] kaduk joins the room
[13:51:55] <wiesand> Pity. But thanks for the notice.
[14:01:09] <wiesand> Hello crowd
[14:02:00] <kaduk> Small crowd, eh?
[14:02:22] <wiesand> Kinda
[14:03:02] <wiesand> Daria, are you out there?
[14:05:11] Marc Dionne joins the room
[14:05:27] <wiesand> It seems not.
[14:06:14] <wiesand> For the record: 1.6.10 blocks on 11517. 1.6.10.1pre1 blocks on 1.6.10.
[14:06:32] <wiesand> Hi Marc, any bad surprises in the final Linux 3.17?
[14:06:51] <Marc Dionne> hi Stephan, not that I know of
[14:08:21] <wiesand> Thanks.
[14:08:38] <wiesand> Ben, anything you want to say about pre-1.8?
[14:09:46] <kaduk> Er, like about the getcwd thing, or about the patch series I'm sitting on that fixes blockers for the 1.8 branch?
[14:10:45] <Marc Dionne> where are the details on getcwd - was in the IRC channel?
[14:11:08] <kaduk> There was a little bit in the IRC channel, and some talk on MIT zephyr.
[14:11:15] <wiesand> Well, any news about the getcwd rumours would help, (is it a 1.6.10.1 showstopper?), but I meant the changes to go in before the 1.8 branch.
[14:11:45] <kaduk> I guess I had sort of mentioned a workaround procedure here that was just copied from the last time we had getcwd issues, and is probably not actually a reproducer.
[14:12:36] <kaduk> IIRC, Anders has reverted 11358 starting from 1.6.10pre1, and is running that on one of the scripts.mit.edu machines, but we don't have a reliable way to reproduce the issue there, so there isn't any real data yet.
[14:13:25] <kaduk> I think that the Scripts folk got a few user support tickets about it, probably from cron jobs. Let me see if I can find anything quickly.
[14:13:51] <shadow@gmail.com/barnowlE5B64A04> ok, does this send?
[14:13:59] <shadow@gmail.com/barnowlE5B64A04> good. yes, i am here
[14:14:02] <wiesand> :)
[14:14:45] <kaduk> Hmm, someone got
shell-init: error retrieving current directory: getcwd: cannot access parent directories: No such file or directory
while logging in to one of the scripts machines.
[14:15:24] <kaduk> But that's basically all the data we have.
[14:15:25] <wiesand> Sounds familiar.
[14:15:39] <wiesand> And it was 1.6.10pre1?
[14:15:57] <kaduk> If it was going to be a blocker, I think it would block 1.6.10, but I'm not convinced that I can suggest that as the right course of action.
[14:16:55] <wiesand> 11358 seems important.
[14:17:15] <wiesand> Sigh.
[14:17:28] <kaduk> I mean, maybe the answer is just "linux is horribly broken forever".
[14:17:55] <wiesand> One way or the other, yes :-(
[14:17:56] <kaduk> But yeah, Anders went through the log of src/LINUX between 1.6.8 and 1.6.10pre1 and that was the only thing that stuck out as a likely cause.
[14:19:18] <wiesand> So we get to choose the way the Linux client is broken. Until it presents the volume mounts as actual mounts to the kernel?
[14:19:29] <kaduk> Something like that...
[14:20:19] <wiesand> Ok, we don't have a report in RT or on the lists. I think we shouldn't block 1.6.10 on this.
[14:20:59] <kaduk> No objection here.
[14:21:16] <wiesand> If the issue does come up (I'll be bitten by it) again, we can try to address it in 1.6.10.1 .
[14:21:43] <wiesand> Daria, any chance you could look at 11517?
[14:22:10] <kaduk> I suppose one could claim that fixing it is just "OS-specific support", appropriate for a patch release ;)
[14:22:14] <wiesand> NB Buildbot did verify PS1, and all that changed in PS2 is the commit message, in the way requested by Jeffrey.
[14:22:55] <kaduk> I'm slightly surprised that buildbot will continue waiting for a buildslave that's offline, but haven't actually looked at our config.
[14:23:58] <wiesand> Has Jason been contacted about this?
[14:24:24] <wiesand> Or: who's in charge of that AIX builder?
[14:24:54] <shadow@gmail.com/barnowlE5B64A04> uh. it's the same machine i had access to, so it's the umu.se machine
[14:25:29] <wiesand> Can you fix it? ;-)
[14:25:58] <kaduk> > Jason
I haven't, and Jeffrey didn't say that he had, so "probably not".
[14:26:27] <kaduk> Kind of unfortunate that buildbot.openafs.org:8010/buildslaves doesn't list the Admin field for slaves that are not currently connected...
[14:26:38] <shadow@gmail.com/barnowlE5B64A04> i think my account expired
[14:26:55] <wiesand> mailing Jason...
[14:27:58] <kaduk> I guess the next topic is 1.8, then?
[14:28:26] <wiesand> Probably.
[14:28:49] <kaduk> I pushed a few more changes yesterday, with the topic 'prebranch', which was a mistake, as none of them actually need to be done before branching, I think.
[14:29:07] <wiesand> (Daria, any reason you approved 11517 but didn't submit it?)
[14:29:15] <shadow@gmail.com/barnowlE5B64A04> jason is currently acting director of his department so he is kinda
snowed under.
[14:29:22] <wiesand> Ah.
[14:29:40] <shadow@gmail.com/barnowlE5B64A04> fixed. sorry
[14:29:42] <kaduk> Probably the most interesting thing is a first cut at a -X option to the fileserver.
I haven't tested it, yet, though :)
[14:30:03] <wiesand> I sent him a mail. Maybe he just didn't notice. Otherwise, well we'll have to wait.
[14:30:27] <wiesand> Daria: Thanks.
[14:30:43] <wiesand> I looked at the -X change. Looks interesting.
[14:31:01] <kaduk> I did go through and re-review the changes that we do think are blockers for branching, and I think I'm happy with them.
[14:32:02] <kaduk> I think that all of the changes to versions that were verified by buildbot are small, so it might be okay to merge them before buildbot starts marking things as verified again, but I can understand if the thought of doing so makes people uncomfortable.
[14:32:16] <kaduk> It's too bad we don't have Simon here today.
[14:33:36] <kaduk> "So, Daria, how do you feel about pushing the button?"
[14:35:13] <wiesand> (pull-up of 11517 is 11530)
[14:35:30] <shadow@gmail.com/barnowlE5B64A04> https://www.youtube.com/watch?v=jdDwxdqLyGg
[14:35:52] <shadow@gmail.com/barnowlE5B64A04> what button is being pushed anyway?
[14:36:13] <wiesand> "This video is not available :-/"
[14:36:34] <kaduk> Bummer, must be US-only.
[14:37:04] <shadow@gmail.com/barnowlE5B64A04> it's the video to "Galvanize". anyway, if you google the lyrics it's
real obvious what i was referencing. but it's not important
[14:37:32] <shadow@gmail.com/barnowlE5B64A04> The time has come to push the button. My finger is on the button.
[14:37:33] <kaduk> I think it would be ... the series up to 11485, possibly excluding 11472?
[14:37:35] <wiesand> Ah, it's a music video, and not a live performance? That explains it. The youtube vs. GEMA war...
[14:40:09] <kaduk> Yes, a music video.
[14:40:54] <wiesand> Anyway, found the lyrics. "Don't hold back" sounds good.
[14:41:16] <shadow@gmail.com/barnowlE5B64A04> so what button do you want me to push
[14:41:43] <wiesand> Ben's probably talking about "Publish & Submit" ?
[14:41:50] <kaduk> The 'merge' buttton, for a pile of ~15 changes.
Do you want the actual list of numbers?
[14:42:45] <wiesand> The list would be useful in any case.
[14:44:54] <shadow@gmail.com/barnowlE5B64A04> if it's a stack give me the bottom and i will just do it
[14:45:22] <shadow@gmail.com/barnowlE5B64A04> what is the first thing?
[14:45:47] <kaduk> 11458 through 11485
[14:46:17] <kaduk> maybe skip 11472.
We also need 11452 and 11453, which are not part of that stack.
[14:50:55] <kaduk> Thanks!
[14:59:03] <wiesand> Ben, you think this is what's required before branching off 1.8?
[14:59:35] <kaduk> I think that that is enough to address all the issues we said were required before branching.
[14:59:59] <kaduk> Maybe I should try to build a debian package from master again once they're all merged, just to double-check.
[15:00:20] <wiesand> I think they are, except for 11472
[15:01:33] <wiesand> If I haven't sent out minutes this time tomorrow, poke me.
[15:02:57] <wiesand> I'll check the individual build slave results for 11530 later, and if all are ok I'll roll new tarballs.
[15:03:23] <kaduk> Sounds reasonable.
[15:03:25] <wiesand> There's a minor update to NEWS in 11531 too.
[15:03:51] <wiesand> But that should then hopefully be 1.6.10, finally.
[15:05:08] <wiesand> 1.6.10.1pre1 should follow swiftly...
[15:05:49] <wiesand> Or do we believe we don't need it?
[15:06:51] <wiesand> We'll see.
[15:06:58] <wiesand> Anything else to discuss today?
[15:07:04] <kaduk> Hmm.
[15:07:12] <wiesand> Hmm?
[15:07:25] <kaduk> "Do we feel empowered to tell Simon to press the 'make a new branch' button (not a single button, of course)?"
[15:08:43] <wiesand> We should give Jeffrey and others a chance to comment, I think.
[15:10:04] <wiesand> That's also why this time I must find the time for the minutes.
[15:10:11] <kaduk> *nods*
[15:10:42] <wiesand> Of course, feel free to send mail to the list(s) on the 1.8 topic independently.
[15:12:43] <wiesand> NB how will we handle pull-ups once we have the new branch? master -> 1_8_x -> 1_6_x ?
[15:13:08] <wiesand> Of course I expect the rate of changes hitting 1.6 to decrease significantly.
[15:13:20] <kaduk> I think pullups can go from master to 1.6 and 1.8 independently.
But yes, the rate would be smaller.
[15:14:16] <wiesand> That would be faster, but bears some risk that we'll have a fix in 1.6 that's missing from 1.8 .
[15:15:14] <wiesand> I trust in the gatekeepers setting the rule here ;-)
[15:15:16] <kaduk> Well, it could work just fine either way.
"Maybe we should talk about it next week."
[15:16:26] <wiesand> Yes.
[15:16:33] <wiesand> SO, more to discuss today?
[15:16:57] <kaduk> I think I'm done.
[15:17:35] <wiesand> You did an impressive amount of great work!
[15:17:54] <kaduk> Thanks.
[15:19:20] <wiesand> Ok. Thanks a lot everyone!
[15:20:00] wiesand leaves the room
[15:27:31] Marc Dionne leaves the room
[15:38:09] <kaduk> (The debian package builds fine based off current master.)
[17:01:08] shadow@gmail.com/barnowlE5B64A04 leaves the room
[17:01:17] shadow@gmail.com/barnowlE5B64A04 joins the room
[17:18:52] <kaduk> Though, we should probably clean up 'make clean'. Lots of stuff left in tests/, some libtool droppings, and some in hcrypto (oops, that's probably my fault).
[17:27:41] shadow@gmail.com/barnowlE5B64A04 joins the room
[17:30:40] shadow@gmail.com/barnowlE5B64A04 leaves the room
[21:35:16] kaduk leaves the room