- Wednesday, July 27, 2016

release-team@conference.openafs.org

Wednesday, July 27, 2016< ^ >

Room Configuration

Room Occupants

GMT+0
[09:00:03] Jeffrey Altman leaves the room
[09:01:10] Jeffrey Altman joins the room
[12:53:42] meffie joins the room
[13:59:04] wiesand joins the room
[14:00:08] mvita joins the room
[14:00:15] <wiesand> Hello
[14:00:51] <meffie> hello
[14:01:38] <wiesand> Let’s start?
[14:01:59] <mvita> hi
[14:02:13] <wiesand> Hi
[14:02:18] <wiesand> So, on Linux 4.7...
[14:02:36] <wiesand> anything besides what we have in/on 12348?
[14:02:41] <mvita> yes, I have some test results
[14:02:50] <wiesand> Ah :)
[14:03:12] <mvita> with Joe's help, I was able to build master openafs on 4.7.0 (GA)
[14:03:22] <mvita> and run without any (new) problems
[14:03:44] <wiesand> good news
[14:03:52] <mvita> in particular, I exercised multiple parallel threads doing reads and writes in AFS
[14:04:00] <mvita> and found no issues
[14:04:06] <mvita> I will continue to test today
[14:04:25] <mvita> I did NOT observe the getcwd problem even though I tried to provoke it
[14:04:41] <mvita> but I believe my cache may not have been small enough
[14:04:52] <mvita> nor had I reduced the frequency on shakeloose
[14:05:00] <mvita> so I will be doing more testing with that today
[14:05:09] <wiesand> Thanks!
[14:05:22] <mvita> s/frequency/period/
[14:05:35] <wiesand> Any news on splice() or "shake harder" ?
[14:05:45] <jhg> not yet
[14:06:06] <wiesand> ok, thanks
[14:06:20] <wiesand> next item on my agenda was 12291 and 12292
[14:06:33] <mvita> stephan, before you go on
[14:06:44] <wiesand> yes?
[14:07:14] <mvita> I did want to mention that we are getting closer to the point of having a linux buildslave that can run against the latest changes as they happen
[14:07:31] <wiesand> That’s good news.
[14:07:36] <kadukoafs@gmail.com/barnowl0226D601> I did not get to look at whether 1.6 differs from master for 1229[12]
[14:07:47] <mvita> so I'm hoping we can use it on the 4.8 release candidates
[14:07:56] <wiesand> Mark: great.
[14:08:18] <wiesand> Ben: I wondered about your comments regarding Solaris 2.6
[14:08:36] <wiesand> Expelling that from master/1.8 is fine. But what about the "stable" 1.6?
[14:09:11] <kadukoafs@gmail.com/barnowl0226D601> *shrug*
[14:10:03] <mvita> oh, I'm just catching up
[14:10:05] <wiesand> Typical Gatekeeper/Guardian decision ;-)
[14:10:21] <mvita> yes please I would like to see 12291 and 2 in 1.6x
[14:10:36] <kadukoafs@gmail.com/barnowl0226D601> It doesn't look like we have version defines before 59, IIRC
[14:10:37] <mvita> if I'm understanding the question on the floor
[14:11:16] <mvita> I can submit those
[14:11:39] <wiesand> The question is "will they break older Solaris versions 1.6.x currently still supports" ?
[14:11:54] <wiesand> (and if so, do we care?)
[14:12:08] <mvita> ah.
[14:12:18] <mvita> well, could be done w/ autoconfig
[14:12:26] <mvita> if that's a real concern
[14:12:42] <mvita> but 2.6 is very long dead
[14:13:04] <mvita> 10yrs since EOS, I believe
[14:13:12] <kadukoafs@gmail.com/barnowl0226D601> Yeah, I don't really think it's worth worrying about.
[14:13:12] <meffie> oh man, 2.6
[14:14:01] <wiesand> Ok, I’ll pull them up (Mark’s time is better spent on stuff I can’t do)
[14:14:51] <jhg> It would be nice to have some sort of quantitative data on kernel versions that AFS users are running these days.
[14:15:33] <kadukoafs@gmail.com/barnowl0226D601> I did ask -info specifically about linux versions a few months back.
[14:15:45] <mvita> I remember that, was just going to mention it
[14:15:51] <wiesand> Yes it would. But those still running 2.6 fileservers won’t tell us on the list, and I object to having our software "calling home".
[14:16:06] <mvita> HOWEVER
[14:16:08] <kadukoafs@gmail.com/barnowl0226D601> There were a couple unicast replies, IIRC, but nothing terribly
shocking.
[14:16:13] <mvita> it is possible to find out
[14:16:20] <jhg> wiesand: it need not call home in order to report such things usefully and anonymously
[14:16:29] <mvita> in some circumstances
[14:17:06] jhg nods
[14:17:58] <wiesand> Ben mentioned that it would likely break the build.
[14:18:29] <kadukoafs@gmail.com/barnowl0226D601> One of them would; the other would not.
[14:18:39] <wiesand> In that case we can just wait for complaints and retrofit an autoconf test if there are.
[14:18:54] <wiesand> We’ll include both, so that’s fine.
[14:19:14] <meffie> sounds good.
[14:19:53] <wiesand> So, do we go for a 1.6.18.3?
[14:20:02] <mvita> yes
[14:20:13] <mvita> in regard to Ander's fix
[14:20:15] <wiesand> With those changes discussed so far, + the OS X candidates?
[14:20:18] <mvita> it is sufficient for build
[14:20:34] <wiesand> ok, thanks.
[14:20:57] <wiesand> as usual I’ll verify it doesn’t break EL5/6/7 before even merging anything
[14:21:02] <mvita> and the concerns about parallel lookup are important but do not seem to matter in 4.7 yet.
[14:21:28] <kadukoafs@gmail.com/barnowl0226D601> It would be nice if a SNA person could throw the OS X 10.11 patch
against an actual 10.11 system; I think we could get the existing
buildslave into the triggered rotation once it passes on master.
[14:21:47] <wiesand> 12344?
[14:21:51] <mvita> This is because OpenAFS is still being called (I believe) with the former exclusive lock instead of shared.
[14:22:08] <mvita> but we will most certainly have to change over in 4.8
[14:22:18] <wiesand> Mark: ah, "good" ;-)
[14:22:25] <wiesand> re 4.8: uch
[14:22:28] <wiesand> ouch
[14:22:45] <wiesand> Doesn’t sound like a small change...
[14:22:46] <mvita> it might not be bad, I have some experiments to perform
[14:23:14] <mvita> Mister VFS, Al Viro, made some testing suggestions to out-of-tree filesystems that I plan to follow
[14:23:29] <kadukoafs@gmail.com/barnowl0226D601> (yes, 12344)
[14:24:39] <wiesand> Is anything missing from the 1.6.19 candidate list I had in the agenda?
[14:25:14] <wiesand> The Ubik fixes need to be sorted out on master first.
[14:26:11] <wiesand> The others seem straightforward to me (arguably, most could be sipped), except the last two (12308 and 12328)
[14:27:01] <meffie> a minor thing spotted yesterday, pulling up the gerrit..
[14:27:45] <meffie> gerrit 10679
[14:28:35] <meffie> and i have some other solaris kstat patches to be pushed to gerrit master
[14:28:55] <mvita> would 10679 be approproriate for 1.6.18.3?
[14:30:03] <kadukoafs@gmail.com/barnowl0226D601> seems like it would
[14:30:49] <meffie> well, it's useful, but not super urgent.
[14:30:50] <wiesand> If someone would jot down a sentence for the release notes...
[14:31:42] <wiesand> ok, the 1st paragraph of the commit message is close to that
[14:31:43] <meffie> ok, i can cherry-pick it for 1.6.x too.
[14:32:12] <wiesand> So no objections to including that in 1.6.8.3.
[14:32:44] <wiesand> But I’d rather not delay .3 for this or more of those not yet even submitted.
[14:32:48] <meffie> i though .3 was for platform fixes, though.
[14:33:12] <meffie> i would not object to waiting for 1.6.19 (with the other kstat patches to be pushed)
[14:33:28] <mvita> kaduk: I didn't see you get an answer about testing 12344
[14:33:45] <mvita> I can ask here
[14:34:02] <mvita> I don't work w/ that myself
[14:34:28] <wiesand> Mike: deferring this to 1.6.19 seems a bit cleaner. But if you want it badly in 1.6.18.3 that’s ok.
[14:34:28] <mvita> but I'm pretty sure we have the test system for that
[14:36:47] <wiesand> In any case added 10679 to the list of things to keep track of.
[14:36:53] <meffie> thanks!
[14:37:25] <wiesand> Shall we aim for getting 1.6.18.3 out within a week from now?
[14:38:14] <mvita> okay
[14:38:31] <wiesand> That would give us some window to work on 1.6.19 before the Linux 4.8 release.
[14:38:31] <meffie> ok
[14:39:40] <wiesand> And I think it’s doable, unless problems are found during testing. Mark: if you find any, please let us know asap
[14:39:50] <mvita> certainly
[14:40:36] <wiesand> I think that concludes the 1.6 part
[14:40:41] <wiesand> On to 1.8 ?
[14:40:59] <mvita> ok
[14:41:48] <kadukoafs@gmail.com/barnowl0226D601> Any plans to do some code review?
[14:42:23] <wiesand> 12344 looks good ;-)
[14:43:01] <wiesand> But seriously, there is a severe shortage in that area.
[14:43:29] <wiesand> On both branches
[14:44:47] <meffie> yes, i am in the middle of code reviews actually. doing 11790 now (well yesterday)
[14:45:37] <meffie> and i have a new patchset for 11978 to push
[14:50:47] <wiesand> solaris mmap pullups are 12349 12350
[14:50:59] <mvita> thanks.
[14:51:07] <meffie> nice. thanks.
[14:52:07] <wiesand> Is there anything else to discuss today?
[14:52:18] <mvita> not from me
[14:52:38] <jhg> negative
[14:52:46] <meffie> is there any news about the <redacted> ?
[14:52:52] <kadukoafs@gmail.com/barnowl0226D601> I don't think so.
[14:53:03] <meffie> will it just be in 1.6.19?
[14:53:44] <wiesand> That would require us to eventually release anything not 1.6.18.x ;-)
[14:53:56] <meffie> heh, yes, i suppose so.
[14:54:32] <meffie> i have some "sort of related fixes"
[14:54:42] <kadukoafs@gmail.com/barnowl0226D601> I think the plan is still to get it in 1.6.19, yeah
[14:55:15] <meffie> that sounds good to me.
[14:55:30] <wiesand> So let’s try to get out 1.6.18.3 quickly (hint: 1.6 changes need review too)
[14:55:30] <meffie> thank you.
[14:56:19] <meffie> ok, thanks have a good week.
[14:56:25] <wiesand> But I’d like to run. Thanks a lot for participating today!
[14:56:36] wiesand leaves the room
[15:35:44] meffie leaves the room
[16:36:46] Jeffrey Altman leaves the room
[16:36:47] Jeffrey Altman joins the room
[18:07:43] mvita leaves the room
[20:43:42] mvita joins the room
[23:35:13] jhg leaves the room