[05:04:22] --- stephan.wiesand has become available [06:10:04] --- squinney has become available [06:12:17] --- meffie has become available [06:58:33] --- ktdreyer has become available [06:58:43] i suppose i should confirm i am actually here, and able to transmit, and that barnowl is not just messing with me [06:58:48] ok [06:59:20] Let me try too... [06:59:29] ...ok [07:00:14] morning/afternoon [07:00:54] Hello [07:00:57] (I'm here! but in another meeting) [07:02:21] --- deason has become available [07:04:07] Let's start. [07:04:45] Thanks for the (ongoing) review efforts! [07:05:33] I merged a first ~10 changes yesterday: http://gerrit.openafs.org/#q,status:merged+project:openafs+branch:openafs-stable-1_6_x,n,z [07:06:11] If you think some of those shouldn't have been merged, please complain. [07:06:34] hooray :) [07:08:21] I also have a list of open changes sorted by the date of pull-up. [07:09:03] Trying to work through it now, oldest first. [07:09:41] Has everyone received the list I sent around some 15 minutes ago? [07:10:00] hm. had to fish it out of spam [07:10:26] I got it [07:10:45] Derrick: Interesting filter... [07:10:57] got it [07:12:10] stephan: my mail is a mess. i have to mark a lot of crap to shadow@gmail as spam which is legit-looking but is something "someone else who thinks they are shadow@gmail subscribed to" [07:12:49] Derrick just +1'ed 8910 - thanks. [07:13:28] if it helps, 9286 in that list is trivial/noncontroversial [07:14:39] That's what I think. But you made it clear last time that I'm still not supposed to merge changes like that one without sufficient review. [07:15:07] yeah yeah, I just meant, for categorization [07:15:23] Of course there's no real definition of "sufficient"... [07:17:20] Like, 8910 now has postiive reviews by Derrick and Andrew. I tend to consider that sufficient if the change looks simple to me. [07:19:18] 9065 is in the same state now. But it still needs to be rebased. Volunteers? [07:20:07] i'm looking. [07:20:17] yes, i'll rebase and push. [07:20:22] Thanks. [07:20:42] i have looked at everything in the email. i either reviewed (or already had) everything. my only questions (and maybe i should remove the +1s there) are the viced->host nat pings [07:22:06] Derrick: Thanks. I think many of those can go in now, in particular all the fs flushall ones. [07:23:13] 9420 may indeed need more discussion. [07:25:05] Derrick: what's your question about it? [07:25:53] i suppose we should stick with the "merge it in pre1, revert if there is no value" plan [07:25:59] in which case, my +1 stays [07:26:26] Andrew, any comments on your last comment in gerrit? [07:27:50] Or on http://gerrit.openafs.org/#q,status:open+project:openafs+branch:openafs-stable-1_6_x+topic:reverse-nat-ping,n,z ? [07:28:08] (if that's related?) [07:28:45] welll... whichever people want to do [07:29:22] the -2'd commits weren't really intended for a release or prerelease as-is; I'd want to rework them a bit [07:29:40] Got that ;-) [07:30:04] I'm just not sure I understand what you're aiming for. [07:32:37] [I think the afscp changes - 9359 and 9973 - are good to go now too] [07:33:57] If there's nothing more to be said about NAT pings today, let's stay with the plan as outlined by Derrick: "merge for pre1, revert if there's no value" [07:34:13] ok [07:34:14] that was the plan we agreed to last time, was it not? [07:34:23] or am i nuts? [07:35:09] No, but since then there were additional comments... [07:36:24] 9430 seems ok to merge. [07:36:46] nothing in them scared me from the pre1 plan. [07:37:27] Ok, lets' stick with it then. [07:38:06] 9434-6 I consider ready to merge as well now. [07:38:47] And 9070 too. [07:39:32] And 9281-3. [07:40:00] (I'm on a slow link today - loading the gerrit pages takes a while) [07:40:46] 9286-7 too. [07:42:02] And 9288 too. Thanks all, that will help get rid of a substantial part of the backlog. [07:42:19] 9282 I kinda wish had another review, but I'm not sure from whom [07:42:55] as well as 9287 [07:43:53] If you're feeling uneasy about them, you should remove your +1 ;-) [07:44:44] well, I mean, I want there to be 3 +1s [07:45:15] if I wait until there are already 2, that seems not productive [07:46:35] Which brings us back to the "what's sufficient" question... any thoughts? [07:47:00] I didn't mean we had to block on it [07:47:43] --- kaduk has become available [07:48:55] just wishes, but I have many wishes [07:49:10] 1.6.2 and 1.6.4/5 were rather conservative regarding what to pick up and what to defer. We can't go on like this forever... [07:49:24] that's why I'm not suggesting blocking on it [07:49:39] if wishes were trees, ... [07:49:48] Well, there is the question of having a truly stable stable release branch, versus having features that go unused and untested for years. [07:49:50] by the way, if anyone would like to look at the modified retry caps for 9571 discussed previously, they are in 10148 [07:50:48] Ben: right. But let's face it: there won't be a next major stable release anytime soon. [07:51:10] I know. So we are pretty much resigned to taking new features. [07:51:26] And it was you who said that the skew between 1.6 and master is already problematic. [07:51:31] it is. [07:52:46] I'm also working to make it worse ... speaking of which, did anyone look at the bozo locking strategy patch I posted that goes on top of Chas's work? [07:53:17] not yet [07:56:23] not yet [07:56:39] not yet :-) [07:56:52] Maybe I should pester jhutz about it. Anyway, it's off-topic for here. [07:56:55] but really really happy to see lwp die some more :) [07:57:06] How about the later submissions to gerrit? [07:57:40] Andrew mentioned 10148. [07:57:43] speaking of bozo, is 10148 correct now? [07:59:03] i can cherry-pick for 1.6.x if the commit on master goes in [07:59:34] Ok. [07:59:45] Looks okay. Let me log in and +1. [08:00:23] And I wanted to ask about 10131 (no uncmpressed tarballs) [08:02:56] May I just merge it? It's trivial, and easy to revert. But it's on page 3 in the list of gerrit open changes for 1.6 and the recent activity. I'd really like to reduce the number of open changes for 1.6 as much and as soon as possible. [08:03:19] you're the boss :) [08:03:21] I don't have a particular tie to uncompressed tarballs. Jeffrey, do you know if folks building on Windows would be affected? I guess if you have something that speaks tar, you're already doing pretty well. [08:04:31] Andrew: yeah, right... [08:04:31] --- shadow@gmail.com/barnowl5ABA57FD has left [08:04:42] --- shadow@gmail.com/barnowl5ABA57FD has become available [08:05:49] Worst case: someone complains and we'll add them back. Unzip, add the MD5 sums. [08:08:01] Regarding bugs to fix in 1.6.6: [08:09:54] One was mentioned by Jeff a while ago. Can't find it quickly right now. [08:10:16] there were two RT tickets [08:10:26] one was the fbsd build thing which is pulled up and on the list [08:10:51] the other was the buserver -servers crash, which was not actually a regression (though it was reported as once) [08:10:51] Anything else that's serious? What about the horror story from Harald? [08:11:04] the latter has a fix I don't think I pulled yet, but I can [08:11:15] Horror story from Harald? [08:11:28] it's not complex and it only affects buserver, so I don't think it's too much to get it in... [08:11:39] File server segfaults. Then the salvager segfaults.... [08:12:01] Ah, that one, right. I think I glanced at the buserver patch and it seemed okay, let me find it again. [08:13:17] I'll have time to compile a list of those open issues this week. Didn't mange today tough :-( [08:13:47] NB Harald's report actually stopped me from commissioning our first 1.6 production server... [08:14:21] the salvager one we have a fix for, and only happens if you '-orphans attach', iirc [08:14:28] it's also not specific to 1.6, I believe [08:14:41] the fileserver one we have no info on, because he hasn't said anything further about it [08:15:26] We haven't had any such issues for many years. And I'm a coward... [08:16:04] Well, my 1.4.14+mumble fileservers are segfaulting once every week or two, and my "solution" is probably going to be "upgrade to 1.6" [08:16:59] (They were 1.4.8 prior to rxkad-k5) [08:17:49] My 1.4.14 (now 1.4.15) ones worked for years. [08:18:39] Well, the "mumble" is some patches that I picked from the tip of openafs-stable-1_4_x that weren't in 1.4.14.1 [08:18:52] Which doesn't mean 1.6 ones wouldn't work. [08:19:21] It's just that Harald's report seems like the only serious one against 1.6 for quite a while. [08:19:25] well, if it helps you, there are indeed many known ways that 1.4.14 viced can crash [08:20:07] I'm sure ;-) [08:20:21] ? because of a single segfault? [08:20:21] did harald open a ticket? does he have a core? [08:20:43] I don't have symbols for my 1.4 binaries but do still have a build tree for the 1.6 packages, which is part of the motivation. Staring at disassembly makes me suspect that we're trying to sprintf something that dereferences a NULL pointer, but I haven't gotten to look through the patch history for anything suspicious. [08:21:30] Mike: no he didn't open a ticket. It's still scary. [08:22:26] well, crashes happen, if we find the reason, it can be fixed. [08:23:09] (The main reason for not deploying the 1.6 server right now was that I'm not immediately available for fixing any fallout for the next few weeks). [08:23:28] speaking of rt, who do i ask for to get an account? [08:23:30] well, we've had segfaults on 1.6 before; I'm not sure if we fixed any for this iteration, but 1.6.3 had segfault fixes [08:23:52] (i could at least clear spam, for example) [08:24:18] or 4, or whatever [08:24:40] I think clearing spam requires more privilege than just an account. (I don't remember if I can clear spam or not, these days.) [08:25:00] Ok, I just wanted to ask whether there's anything going on w.r.t. that report that should be considered for 1.6.6. I take it there's not, at least for the fileserver part. Fine. [08:25:24] kaduk, stephan; removing uncompressed tar files for Windows users is not an issue. [08:25:57] Ah, thanks. Let's remove them. [08:26:16] stephen: haba made noises to the list and then disappeared, his servers are all upgraded and he has said nothing since. all requests for info were responded to with "I have no time". I would ignore him. [08:26:31] Uploading those from here is a real nuisance too. They take longer tan all the rest together. [08:27:03] there is a fix for his salvager issue, which could go in [08:27:04] in order to build on windows you need cygwin and cygwin has both tar and gunzip [08:27:12] meffie, didn't i promote yours to an account a while ago? [08:27:35] i have the power. lemme look [08:28:23] ah. nope. which email should i promote to a real account? you have 2 [08:29:06] oh, ok, the the sna one would work. [08:29:26] Finally found 10115 for the buserver crash; it looks like there's no pullup submitted. Andrew, were you planning to submit one? [08:30:11] https://rt.central.org/rt/Ticket/Display.html?id=131716 is a new 1.6 ticket [08:32:46] stephan.wiesand: 10165 is haba's salvager thing, if you wanted [08:32:52] kaduk: yeah, I'll get it, one moment [08:34:23] Andrew, thanks. It's at least a candidate for 1.6.6 IMO. [08:34:44] 10166 [08:35:03] I'm not sure I get RT#131716 [08:36:20] But will put it on the list. [08:38:07] Anything else to discuss today? [08:39:51] I'll write up minutes tomorrow. [08:40:07] Thanks a lot everyone. [08:40:17] thank you [08:40:40] have a good day [08:40:52] thank you [08:40:59] Bye [08:41:01] --- stephan.wiesand has left [08:49:48] --- kaduk has left [09:02:15] --- squinney has left [11:57:48] --- meffie has left [12:36:52] --- ktdreyer has left [12:44:53] --- meffie has become available [12:45:23] --- meffie has left [16:02:59] --- deason has left [16:31:12] --- meffie has become available [16:31:32] --- meffie has left [19:49:30] --- Jeffrey Altman has left: Disconnected [19:54:15] --- Jeffrey Altman has become available