16:58:54 #startmeeting network team meeting, 23 April 16:58:54 Meeting started Mon Apr 23 16:58:54 2018 UTC. The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:58:54 Useful Commands: #action #agreed #help #info #idea #link #topic. 16:58:58 ehlo 16:59:19 hi everyone 16:59:22 hi 16:59:23 pad is at https://pad.riseup.net/p/kNjIkulFxsI6 16:59:38 hi everyone !! 16:59:56 hi all :) mostly lurking today 17:00:00 so: **we have 3 weeks left till the 0.3.4 freeze** 17:00:15 pastly: nice with open sourcing the scanner! 17:00:24 how are we doing on the 0.3.4 items on the roadmap? 17:00:43 ahf: asn: thanks :) 17:00:44 * ahf is testing the easy one of his 0.3.4 roadmap items now 17:00:48 everybody look what your name is listed near on the 0.3.4 list: is there anything that might take longer than 3 weeks? 17:00:53 oi! 17:02:21 yes my name is in more tickets than i can probably handle. i will triage this week. most of them are just the result of moving triaging stuff out of 033 . 17:02:25 pastly: cheers! *crosses fingers* 17:02:32 (im not talking about roadmap tickets) 17:02:33 could everybody please use the new "status" column on the roadmap to note what looks happy/sad from your pov 17:02:36 ? 17:02:42 sure 17:04:04 grrr the pad will not connect for me 17:04:09 mikeperr1: yo around? 17:04:23 I am still juggling #25546 subtickets. I might like some dgoulet time to discuss #25883 17:04:55 mikeperr1: i marked two of the roadmap items as "good progress". particularly prop#291 and the simulator. 17:04:59 isis: where are we with wide creates? I know you've been doing some work there... 17:05:20 mikeperry: ping me anytime for this 17:05:28 nickm: there's a "status" column? i only see two "notes" columns 17:05:46 catalyst: maybe it's another tab? 17:05:53 catalyst: we are in the second tab "Roadmap March-Nov 2018" 17:05:53 I'm looking at the "Roadmap March-Nov 2018" tab? 17:05:54 catalyst: it's the roadmap march-nov 2018 17:06:03 tab name ddos 17:06:20 nickm: sorry, nevermind. was looking at ticket review assignments 17:06:47 mikeperry: btw, #25668 is named "investigate 2-guard proposal" and we are definitely investigating. do you think we are good in progress terms? 17:06:56 nickm: i've got encoding and decoding functions and i need to write tests 17:07:11 ok 17:07:27 nickm: do you know how much progress has been made on the other wide create cell tickets 17:07:30 ? 17:07:39 basically none afaik 17:07:47 has anybody else done any of the wide create stuff? 17:07:50 nickm, isis, catalyst: the wide CREATE cell is a bit deprioritize for me as there are 4 names on that ... so apart from that, I'm still confident on my tasks. 17:07:57 I'm afraid I've been 100% on CPU-reduction 17:08:00 (that answers it for me ^ :) 17:08:27 with a plan to _try_ to get to wide-create, but let it sleep from 0.3.4 if we need to do so in order to get everything else in. 17:08:45 asn: sort of. I feel like we're still kind of undecided on a lot of things. I think we've got enough to move forward on #25870 for vanguards.. 17:08:47 where can I get a bit more context on the wide CREATE cells? a spec/roadmap/etc.? 17:09:10 dmr: prop#249 17:09:35 geez, finally, the pad HS address works but not the clearnet one 17:09:43 The "basic modularization preparation" item also seems not to be moving very much. 17:10:03 nickm: not sure we need it afterall actually... 17:10:04 and whether or not the privcount blinding and encryption get in, seems really dependent on how teor4's project is going 17:10:04 nickm: thanks! 17:10:14 dgoulet: "it" ? 17:10:31 isn't the basic modularization stuff a bit related to what dgoulet have started with? 17:10:34 nickm: the ticket... I can re-assess after the meeting what exactly the action items there but I'm not sure there is much 17:10:34 Those seem like the 3 that I think are most likely not to be done by schedule. 17:10:38 being able to compile tor without some components? 17:10:46 or is this a more planning task? 17:10:54 planning task I think 17:10:56 that could be part of it; I think we had envisioned more planning and refactoring 17:11:01 ack 17:11:30 planning happened before I started with #25610 so maybe it could be closed tbh 17:11:50 (#25498 to be closed I meant) 17:12:18 hm. I'm not so sure there -- we listed #25498 as sponsor 3, but #25494 as sponsor 8. 17:12:36 we should see how much we can do there 17:12:39 weird... 17:12:56 odd 17:13:09 the focus on #25494 is related to binary size as well (and thus download size) 17:13:14 maybe that is why? 17:13:18 yeah and it is the main roadmap item 17:13:36 so we should probably label #25498 S8 17:13:36 ahf: yeah that's why #25494 makes more sense as sponsor8 to me 17:13:46 yeah, i agree 17:13:57 Okay. Let's consider that, but IF we do... 17:14:06 we should check in with isabela about how this affects our planning 17:14:19 especially for sponsor3 17:14:30 and consider whether we want to do anything additional for #25498 under s8 other than what we already have planned 17:14:49 (I think isabela is busy atm with blog stuff; so let's check in with her later) 17:14:56 well I personally think that ticket is done, regardless of sponsor but maybe we need some texts for a report there? 17:15:01 (we can check that after) 17:15:15 remind me what s3 is? 17:15:18 :/ 17:15:31 it's one of the NSF things 17:15:37 it's the one with the complicated history 17:15:37 ack 17:15:53 next up is reviewer assignments? 17:16:15 [Note: The tasks I marked in red are not _necessarily_ going to slip, but they currently look like they could, and we know it.] 17:16:28 looks like everybody has <= 2 17:16:35 (re:sponsor8 i will follow up later w/ yall) 17:16:46 does everything look okay there? everybody good to get this done by end-of-week? 17:17:23 mine looks fine. an older bug that wasn't done was marked as done though in the sheet 17:17:37 which one? 17:17:59 i don't really understand how modularity is tied into s3 but that seems like a digression and also idk if i need to know 17:18:10 For rotations, we have isis=bug-triage, community=mikeperry, coverity=nickm, CI=ahf. 17:18:18 i readded it. the #25140 17:18:24 nickm: ack 17:18:28 \o/ heck yeah bug triage 17:18:46 I had some notes from doing CI last week, but they'll come up during discussion time 17:19:01 and on to discussion time, not quite in order: 17:19:02 i have a question on CI: i think our jenkins bots are still a bit sad, should i try to spend some time on that this week? 17:19:16 Let's talk about that 17:19:18 ack 17:19:30 So, I got nearly all the jenkins builds building last week. 17:19:54 right now the only thing that looks unhappy from "tor" in jenkins is "032-extra-arm" 17:20:20 but that one is failing with "Timeout, server build-arm-02.torproject.org not responding" on one of the builders, so I think it's a false positive. 17:20:27 I think we're entering the week with Jenkins in good shape 17:20:39 ok, cool! 17:20:48 metrics-lib-master and stem-ci-linux are failing, but those aren't ours... 17:21:12 yeah 17:21:34 and the freebsd ones that are failing are failing with "packet_write_wait: connection to 81.x.x.x port 22: Broken pipe". 17:21:38 So that's not our code either 17:21:48 hm, i see 17:21:49 This is a change since last week, though: 17:22:07 all through last week, I was finding and fixing jenkins issues related to stuff that we were testing in jenkins only, and which had been broken for a bit 17:22:10 most notably: 17:22:11 windows 17:22:13 32-bit behavior 17:22:23 "make distcheck" 17:22:36 Is this all stuff that we should try to get travis checking too? 17:22:54 do cross-build of win32 with travis? 17:22:55 where by "travis" I guess I mean "something integrated with github 17:23:32 I think that whether we do or we don't, we need to plan to keep checking jenkins and fixing real jenkins failures. 17:23:41 i think that's what the appveyor ticket is about? 17:23:46 either indefinitely, or until our github-integrated thing can do everything 17:23:55 #25549 17:24:19 I'm asking because of isis's question on #25814 -- 17:25:02 Is there anything that we _don't_ want to add to travis, under the theory "jenkins already does that well enough"? 17:26:22 i think whenever feasible we should move stuff off of jenkins and onto stuff that has more affordances such as travis provides 17:26:56 +1 17:27:13 Does that mean we should add "debian packages" to travis? Or keep checking those on jenkins? 17:27:45 we can probably build 64-bit debian packages on travis right? 17:27:59 Not sure :) 17:28:10 plus i really like shifting the responsibility for causing/creating issues to each developer (it sucks to find out after my code is merged that there was a problem somewhere) 17:28:24 isis: +1 17:28:35 wont building debian packages for debian require us to do a debootstrap from the ubuntu into a debian environment before we can build them? 17:28:40 I think I agree, but I also think we should do two more things: 17:28:48 - We should keep looking at jenkins failures 17:28:49 i had the impression that we are bit limited on how much time we have on travis 17:29:15 - We should prioritize the configurations where we tend to find bugs. 17:29:35 plausible? 17:29:53 ahf: I don't actually know how limited we are in the number of configurations we run 17:30:28 nickm: each sub-job has an independent timeout 17:30:35 or that's what i recall anyway 17:30:39 there 17:30:42 sounds good to me. it's good the jenkins bots also writes on IRC when there is failures 17:30:51 there might be an aggregate timeout too, but i'm not sure 17:31:32 ok. it seems like we have rough-ish consensus to move forward with Doing More In Travis and to keep Checking On Jenkins. 17:31:32 we use jenkins for continuous deployment for things that aren't core tor and it all goes to #tor-bots right? 17:31:35 Great. 17:31:56 catalyst: i think so, #bots-bots have a bit more activity than the ci channel 17:32:03 err, #tor-bots 17:32:14 next question: did anything new turn up for 033-stable? Is there anything for us to do on that other than to keep working on bugs and wait for more bugs to be reported? 17:32:54 is 033-stable LTS btw? 17:33:12 We haven't decided what the next LTS should be 17:33:20 i've switched my relay over to the 033 rc and haven't found anything there 17:33:32 #25870 should go in to 0.3.3 I think. if we like it 17:34:10 i might ask for some help on making a good regression test case for #25245 - it has been a bit of a pain to reproduce 17:34:12 I think maybe no on that one -- vanguards are experimental, yeah? 17:34:31 nickm: yeah but if people use them in 0.3.3, it will cause warns on relays. 17:34:58 hm. At the least, let's see how big the patch turns out to be, and try it in 0.3.4 first. 17:35:39 ok 17:36:33 for 034-proposed, I see only 2 tickets listed: #25883 and #25895 17:36:53 oh wait 17:36:57 the name is 034-roadmap-proposed, right? 17:37:14 yes, there are 6 of those. 17:37:41 https://trac.torproject.org/projects/tor/query?status=accepted&status=assigned&status=merge_ready&status=needs_information&status=needs_review&status=needs_revision&status=new&status=reopened&keywords=~034-roadmap-proposed&group=status&col=id&col=summary&col=owner&col=type&col=priority&col=component&col=version&order=priority 17:37:46 ^ a list of all 8 17:38:19 Is it time to go over these and see what we can try to get done before the deadline? 17:38:31 I think I already approved #25756 17:38:36 and #25895 seems important 17:39:20 i doubt we can fit #20212 in. we dont have a plan yet or time to test it. :X 17:39:33 ack 17:39:34 i have looked a bit at windows cross compiling before and could take it, but i'm not super hardcore at the rust toolchain stuff. might need some help there 17:39:48 for #25895 17:40:00 ahf: thanks! it might make sense to look at that after we do/don't merge the libtool patch. 17:40:42 yes, for sure 17:40:50 i'll assign myself to it for now 17:43:22 mikeperry: for #25883 we would need a sketch for a new control-port feature: right now I don't think STREAM events support any kind of exit-side streams; maybe a new controller event type is in order 17:44:15 mikeperry: we can sketch something out some time? I'd be happy to walk though the help/tips if you want 17:44:18 I think I know about how to do that 17:44:49 pfew! 17:45:05 That was a lot of stuff to look at. I hope everybody's doing okay! :) 17:45:12 Anything else we want to talk about here? 17:46:09 isis: poking rust people about the bool thing? who was going to do that? i'm not sure 17:46:26 catalyst: ah, i was going to do that, sorry 17:46:34 nickm: the main thing I actually need is just a mapping from onion service stream to circ id, and then bandwidth events for those. 17:46:41 catalyst: i have an open tab with a half-written comment 17:46:46 catalyst: unless you wanted to 17:46:51 isis: cool, thanks! 17:47:30 isis: i.e. please go ahead with your comment :) sorry for ambiguity 17:47:57 okay. Hearing no new topics come up, I'm going to call the meeting done! 17:47:59 Thanks, everybody! 17:48:27 o/ 17:51:31 nickm: might want to #endmeeting before your forget :) 17:52:05 #endmeeting