[01:57:07] --- wiesand has become available [01:57:25] test [04:32:54] --- wiesand73543 has become available [04:32:54] --- wiesand73543 has left [04:32:55] --- wiesand73543 has become available [04:32:55] --- wiesand73543 is now known as wiesand13379 [04:32:55] --- wiesand13379 has left [04:32:55] --- wiesand13379 has become available [04:32:55] --- wiesand13379 has left [04:32:55] --- wiesand6270 has become available [04:32:55] --- wiesand6270 has left [04:32:55] --- wiesand48358 has become available [04:32:55] --- wiesand48358 has left [04:32:55] --- wiesand51570 has become available [04:33:25] --- wiesand51570 has left [04:33:36] --- wiesand19402 has become available [04:33:36] --- wiesand has left [04:40:27] --- meffie has become available [04:47:28] Hi. Do things look normal to you? [04:49:20] --- wiesand19402 has left [04:49:30] --- wiesand has become available [06:58:30] --- deason has become available [07:01:33] hi [07:01:50] Hi [07:01:58] hello hello [07:02:48] No Marc today :( [07:02:52] Howdy. [07:02:59] hello [07:03:15] Thanks for being here. [07:03:19] --- Marc Dionne has become available [07:03:26] --- Simon Wilkinson has become available [07:03:32] Ah, Hi Marc :) [07:03:39] Hello Simon [07:03:50] hi [07:03:52] Hi, sorry a bit late [07:04:22] Shall we start with 1.6.7? [07:04:49] We seem to be missing Daria [07:04:50] sure [07:04:54] Ah, no. [07:04:56] we are? [07:04:59] So, 1.6.7 [07:05:16] Simon, you take over the chair? [07:05:18] I have a tree ready to push for master. Stephan has a tree ready to push for 1.6. [07:05:34] Release announcement and security advisory are in the -announce queue [07:05:34] i am unembargoing the afs volume now [07:05:47] Do we have stuff ready to push to the website? [07:05:52] then i will release the volume and its parent; push the web pages to gerrit, and fire all the queues [07:06:24] the volume has no replica yet. [07:06:41] 1.6.7 not replicated. sw.openafs released. [07:06:51] 10994 and 10995 in gerrit [07:06:58] Okay. I'll push the merge to master. [07:07:33] I push to 1.6.x now? [07:07:45] go ahead. the volume is visible [07:07:50] well, up to simon [07:08:44] Just waiting for gerrit to accept the master change... [07:08:51] fatal: The remote end hung up unexpectedly [07:08:59] what are you psuhing to? [07:09:14] git push origin HEAD:refs/heads/openafs-stable-1_6_x [07:09:30] and you have origin set to gerrit (port 29418)? [07:09:54] web site firing [07:10:11] I should. How to verify? [07:10:17] git config -l [07:10:18] 'git remote -v' [07:10:24] er, that [07:10:37] Well, I just messed that up. Ooops. [07:10:44] origingit://git.openafs.org/openafs.git (fetch) origingit://git.openafs.org/openafs.git (push) [07:10:52] yeah, that won't work [07:11:06] git remote add gerrit ssh://gerrit.openafs.org:29418/openafs [07:11:23] git push gerrit HEAD:refs/heads/... [07:11:49] the git protocol thing is not pushable to. it's only for fetching. [07:11:58] on its way [07:12:02] Yeah, make sure and push to refs/heads/. That's how I messed up. [07:12:05] looks ok [07:12:17] web site should be visible. if anyone wishes to confirm, please do. then i will fire the announcements [07:12:25] I screw up refs/heads vs. refs/for about half the time when updating the wiki :-/ [07:12:41] Recent News: security advisory 2014-001 [07:13:09] Daria: Could you tag 94ffd115f0510fde86339395f88d41d9d56950f4 ? [07:13:21] release announcement sent first so it gets the nuber i used in the web page [07:15:32] i see the website update. [07:16:02] security issue announced. [07:16:02] main page say "read the 1.6.6 release notes" [07:16:06] tag pushed [07:16:16] the link works though [07:16:29] Cool. I just need to send the announcement to oss-security [07:16:44] release page looks good [07:17:18] And we can now sit back and wait for the howls of complaint [07:17:49] typo corrected [07:18:01] web site refreshed. [07:18:33] Asked Stephen to lift the acl on his 1.6.7 build dir so I can rsync today's fedora moules [07:18:42] ok [07:18:59] And that's probably it. Great! [07:19:38] That seemed pretty painless, actually. This is the way to do them in future? [07:20:02] Why not. [07:21:09] whew, finally i can talk again. i hate keeping secrets :) [07:21:36] "Normal" procedures with limited visibility to the public would be even better. But it works this way. [07:22:06] Mike: yeah. [07:22:44] Thanks a lot to all who helped with 1.6.7! [07:22:50] Back to the agenda? [07:23:07] Marc: Any Linux news? [07:23:32] we're still ok as of yesterday's mainline, which is still in the merge window, but probably at the tail end of it [07:23:48] A glimpse of hope... thanks. [07:24:12] On to "problem reports" then. [07:24:36] I found 10964 for the stack overflows. Is there another one? [07:26:22] I think that's the only one that made it to gerrit (other than the original thing that Simon -1'd) [07:26:42] Is it for 1.6.8? [07:28:12] i'm currently doing more of the same, but with the treq. [07:29:04] I'd be inclined to land all of the stack reduction stuff at once. Whilst 10964 will help in general, it won't help with the specific issue that MS raised at the conference [07:29:08] at the moment, just verifying the sizes with chas's stack_usage.pl [07:29:26] Waiting for the whole package before we decide whether/when to pull up makes sense to me. [07:29:32] yes [07:29:37] Fine. [07:29:50] The other problem I had in the agenda isn't present on 1.6.x [07:30:02] Missed that yesterday, sorry. [07:30:22] --- wiesand9640 has become available [07:30:22] --- wiesand9640 is now known as wiesand94612 [07:30:22] --- wiesand94612 has left [07:30:22] --- wiesand94612 has become available [07:30:22] --- wiesand94612 is now known as wiesand66070 [07:30:22] --- wiesand66070 has left [07:30:22] --- wiesand66070 has become available [07:30:22] --- wiesand66070 is now known as wiesand39732 [07:30:22] --- wiesand39732 has left [07:30:23] --- wiesand39732 has become available [07:30:23] --- wiesand39732 has left [07:30:23] --- wiesand68002 has become available [07:30:23] --- wiesand68002 has left [07:30:23] --- wiesand88155 has become available [07:30:53] Were we all kicked out? [07:30:53] --- wiesand has left: Lost connection [07:31:01] i am here [07:31:12] i am here [07:31:24] i am here [07:31:25] still here [07:31:53] still here [07:32:09] Ok, it seems it was just me again :-( [07:32:40] Anyway, I think we're done with "problem reports", unless I missed anything. [07:33:22] So, 1.6.8 (pre2). [07:33:54] Ben, do you want 10987 in 1.6.8? [07:34:34] Yes, please. [07:34:45] I noticed that I was still carrying that locally when I did the pre1 builds. [07:34:57] Ah ;-) [07:35:10] I guess there are no objections? It's FBSD only. [07:35:30] i like the commit message. [07:36:03] do it [07:36:08] I don't get the commit message. But it's not supposed to be different than on master, so... [07:37:06] merged. [07:37:14] The others I guess are more controversial. [07:37:23] 10984 afs: Raise fake free space reporting [07:38:00] it's a bullshit number anyway. using a different bullshit number is still bullshit [07:38:06] so. don't care [07:38:33] I like the change. But I think it's not for a pre2. And I'm sure some will say it's not for 1.6 at all. [07:39:08] I don't see why this would be prohibited from a stable release [07:39:09] I think it's fine for 1.6, but would agree that it's not for pre2 [07:39:25] Ok. Fodder for 1.6.9. [07:39:48] And then the hot threads thing... 10957. [07:40:10] Certainly not for a pre2 either. [07:40:29] But how about 1.6.9? If there's data that it helps? [07:40:32] Not for pre2. Debatable for 1.6, as it does cause some fairly significant changes under the hood [07:41:30] I don't have any specific numbers for OpenAFS. YFS certainly saw a speedup by changing, but we'd fixed a number of other issues before we ran into that one [07:41:32] i assumed it was for a 1.8. [07:42:14] Agreed. I'll try it. If it helps *a lot*, I'll bring it up again. Otherwise: yes, rather for 1.8. Or whatever. [07:42:22] ok [07:42:59] Anything else we'd dearly miss in 1.6.8 at this point? [07:43:26] are we still delaying 9407? [07:43:46] or if that will not be for 1.6 at all I can get rid of it [07:44:41] It does no harm to keep it around. But I'd like to have the option for some more time. [07:44:49] So: yes :-) [07:44:52] okay [07:45:05] just checking :) [07:46:09] so if we're done with 1.6.8.... [07:46:23] the source code patch in the 1.6.7 announcement doesn't apply to 1.4.15 [07:46:27] I think we are. pre2 early next week ok? [07:47:02] since 1.4 eol has still not yet occurred, I thought we'd be releasing patches that do apply to it [07:47:16] pre2 early next week is fine here. [07:47:33] I can send one somewhere; it's just that the older conditional '>' check is still in the way; not complex [07:47:34] There should be a 1.4.15 patch [07:47:51] If you have one already, that would be great and we can link to it from the website [07:48:03] It's in RT. [07:48:13] it's in 131803 [07:49:35] Andrew, could you submit a change to openafs-web? [07:50:47] I don't actually have the patch in 131803; I assume what I have locally is the same, but... [07:50:51] Stephen opened up the 1.6.7 build dir rsync running... [07:52:27] the one i put in 131803 is just: - if (statsVersion > STATS64_VERSION) + if (statsVersion != STATS64_VERSION) [07:53:32] you need to set 'code'; I'll submit something, one moment [07:53:33] Right. And the rx.c patch applies to 1.4.15 as is. [07:54:01] Less worried about the rx.c change. It's not a security issue per-se, just a denial of service. And we have lots of them ... [07:54:11] True. [07:58:14] gerrit 11001 [08:00:07] Hmm, quite different :-/ [08:00:11] sigh. whitespace. [08:00:48] ? it's the same as the 1.6 one, it just removes the lines that are there in 1.4 [08:01:15] yes. the 1.6 one also has whitespace errors. oh, you meant to stephan [08:01:15] It's quite different from the 1.4 patch in the ticket. [08:01:37] it a diff of a diff, so it looks different. thanks for spotting the missing code = EINVAL [08:01:39] web site has it now [08:01:50] any whitespace errors in that are coming from the context in the patch, as far as I see [08:02:36] Yeah, 1.4 doesn't have the whitespace cleanup [08:04:16] oh, it's not a diff of a diff. sorry. [08:04:48] well, in gerrit/gitweb you see a diff of a diff sorta, that's true [08:04:59] Hmm, I better tell Russ... [08:05:04] The only difference between the patch in the ticket and the one in gerrit is the code = EINVAL setting. [08:05:08] (i was looking at unified, that that makes it look like a diff of a diff) [08:05:21] should I send something to -announce? the website has the link, but the -announce email doesn't [08:05:40] I wouldn't bother unless folk ask. It's linked from the website. [08:06:04] WRT Russ, setting code to EINVAL isn't actually necessary - either change protects against the attack. [08:06:43] Ah, good news. [08:07:20] Any negative effects of the missing code=EINVAL? [08:08:00] it looks like you'd return success but no actual statistics [08:08:08] We'll return success with an empty array, rather than an abort packet [08:08:33] which is just potentially confusing to clients; I don't know if they handle that correctly, but no big deal [08:09:04] clients should never hit this, as they should all be sending the right version. Only folk who will hit it are people trying to exploit the bug [08:09:17] yes [08:09:18] If the client is trying to catch the fileserver, they should not be indignant if they get an empty reply. [08:09:19] "breaks exploiters". poor babies [08:10:12] at any rate, thank you for catching that andrew. i missed that. [08:10:30] (at least on 1.4.x) [08:10:56] not "only" exploiters; it's a version argument so you can have different versions [08:11:12] future changes could modify it, or people that changed it themselves or something [08:11:32] yep. [08:11:35] but "not a concern" [08:11:42] moving on? [08:12:03] Okay, let's move on. Thanks for the explanations though. [08:12:11] And I notified Russ. [08:12:30] So: "Testing". Any news? [08:13:17] It seems not. [08:13:34] So, last item: "Next stable branch(es)". [08:13:35] not for corralling up a group of specific testers; I have some things I need to get done in front of it [08:13:43] but I should be able to soon [08:13:56] Ok, let's see next week. It's late anyway. [08:14:30] But since we have Simon and Jeffrey with us today, discussing the next branches seems expedient. [08:15:43] I like Daria's idea to keep "Windows" and "Portable" on independent schedules. [08:15:58] Anyone here who doesn't? [08:16:29] Anyone else who does? [08:16:46] that seems fine, but it's not answering everything :) [08:17:18] I like the idea of saying "even is Unix, odd is Windows" [08:17:24] Not least because Windows is odd :) [08:17:28] heh [08:18:57] it seems really confusing for some users, but it maybe wouldn't be a problem if building for the 'wrong' platform was prohibited or something [08:19:35] It's been like this for a couple of years. [08:19:58] 1.6 for windows was legacy even before it stopped. [08:21:42] Who's going to take the decision in the end? Gatekeepers? Elders? [08:22:15] Gatekeepers, I guess [08:23:34] My preference: 1.8 next stable *x, 1.9 next stable Windows, anything else discussed once those fly [08:23:48] But again, I'm just the 1.6 release manager ;-) [08:24:04] That would be my preference too, but I have no standing in this debate :) [08:24:09] Keep that up, and you might find yourself the 1.8 release manager as well :) [08:24:26] i don't either. i'm sitting [08:24:31] Sounds like fun :-) [08:24:56] personally i think something more obviously different than odd/even would be better. i think that's what we heard from people at the conference as well [08:25:14] i agree. i think the odd/even thing is just not simple/obvious enough [08:25:24] that is my opinion too, but I also don't see any "serious" problems with how things have been going [08:25:46] but that's just from my perspective; so I don't have any motivation for really trying to do anything different [08:27:05] things like strawberry perl and k4w seem pretty common, but eventually you'll have a version number clash and then things get confusing again [08:27:36] KfW, please; k4w makes me think of krb4... [08:27:46] as developers most of us probably adapt to whatever the scheme is, and don't tend to care too much. but some users at CERN seemed to care; a little at least [08:27:52] whoops yeah, that's what I meant, hah [08:28:23] I'm wondering why though. [08:28:28] but yeah other schemes will have their own problems.. [08:28:42] some of the questions seemed to be framed with the assumption that the version numbers would be going back to just one track [08:28:43] We have this occasional "I build 1.7 on Linux and it doesn't work" report. [08:28:48] But that's it. [08:28:50] because I think that's what I lot of people assumed; it's what I thought [08:29:03] so some were just like "when is it going to happen" [08:29:11] but maybe it's not a problem if it explicitly never does happen [08:29:38] I don't see how it can happen within the current model. The frequency of Windows releases and Unix releases just don't match up. [08:29:51] but it seems like it needs input from the community directly [08:30:35] I'm not suggesting it, but you can just 1.8.2 then 1.8.2.1 for win and 1.8.2.2 etc [08:31:21] or you have 1.8.2 for everyone, then 1.8.3-1.8.5 just for windows [08:32:14] but I guess I personally don't see the major problems with the current even/odd way, so I can't really make an informed argument [08:34:28] I guess there's no more input right now. [08:34:49] it really seems like it needs to be brought up with users [08:35:12] since "user confusion" is the main problem/complaint [08:35:18] We can try. I expect 99% not to care at all. [08:36:08] if they don't care then it doesn't matter what decision is made :) [08:36:30] Exactly my point. And as a user, I'm one of them. [08:38:29] I'll try to summarize this in the minutes. Please correct me if you have to, because I'm not sure I get all the ideas. [08:38:40] ok [08:39:10] And the minutes going to -devel may get us some user input. Not from the confused ones though, probably. [08:39:28] So, anything else to discuss today? [08:40:43] i have nothing [08:41:06] Ok. Thanks a lot everyone for this long and productive meeting! [08:41:36] Going offline for an hour. Bye. [08:41:49] --- deason has left [08:41:51] --- wiesand88155 has left [08:42:27] --- meffie has left [09:06:29] --- Marc Dionne has left [12:24:33] Oh, I guess I didn't mention that the commit message is a reference to this quote: http://hitchhikerguidetothegalaxy.blogspot.com/2006/04/beware-of-leopard-douglas-adams-quote.html Though that page may not have enough context for it to fully make sense. [18:04:10] --- Simon Wilkinson has left [18:24:38] --- Simon Wilkinson has become available [20:03:02] --- Simon Wilkinson has left [20:06:26] --- Simon Wilkinson has become available