14:32:48 <karsten> #startmeeting metrics team
14:32:48 <MeetBot> Meeting started Thu Jul 27 14:32:48 2017 UTC.  The chair is karsten. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:32:48 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
14:33:02 <karsten> (sorry to interrupt, nickm, asn, dgoulet!)
14:33:22 <dgoulet> np
14:33:52 <karsten> https://storm.torproject.org/shared/Ou-1QRctynWbF4yedi-MfDsjImFMFSIEP20fbVGCPRa <- agenda pad
14:35:15 <karsten> iwakeh: alright, shall we start?
14:35:23 <iwakeh> yep.
14:35:32 <karsten> * Onionoo backend
14:35:38 <karsten> I hear we have a new host.
14:35:46 <karsten> I didn't log in yet.
14:36:02 <iwakeh> ah, ok.
14:36:09 <karsten> it's supposed to be a second back-end host that runs the hourly updater.
14:36:36 <karsten> it should ideally produce the exact same data. but that might not always work out.
14:37:04 <karsten> still, a good opportunity to remove any inconsistencies between two onionoo instances.
14:37:08 <iwakeh> well, we had differences before.
14:37:11 <karsten> and a big plus in robustness.
14:37:20 <iwakeh> yes.
14:37:33 <karsten> yes, I think there are open tickets for differences.
14:37:51 <karsten> but, not the end of the world if something's slightly off.
14:38:24 <karsten> okay, how do we set this up? would you want to give that a try some time next week?
14:38:25 <iwakeh> no, another backend is an improvement.
14:38:29 <iwakeh> regardless.
14:38:35 <karsten> agreed.
14:38:52 <iwakeh> sure, I can give it a try.
14:39:10 <karsten> cool! do you already have access to omeiense?
14:39:21 <iwakeh> not that I'm aware of.
14:40:00 <karsten> ah, indeed.
14:40:05 <karsten> so, we need to fix that first.
14:40:08 <karsten> I'll create a ticket.
14:40:15 <iwakeh> ok.
14:41:05 <karsten> I think step 1 would then be to log into the new host and set up a new onionoo instance, without any historic data.
14:41:27 <karsten> and maybe even keep that running for a day to figure out whether there are any issues.
14:41:36 <iwakeh> fine, I'll start a list about operation like for CollecTor.
14:41:52 <karsten> the next step would be to copy over the status/ and out/ directories from the primary instance.
14:41:57 <karsten> (starting a list sounds good!)
14:42:02 <iwakeh> well, it's only updates yet for collector.
14:42:31 <iwakeh> shouldn't these be copied first?
14:42:34 <karsten> in order to copy them, we'll have to stop the hourly updater, create tarballs, restart the hourly updater, copy over tarballs, and get it running on the other side.
14:42:44 <karsten> the issue is that they age quite fast.
14:42:58 <karsten> if it takes a few days to figure out any issues, we'll have to copy it once again.
14:43:05 <karsten> 72 hours in collector's recent/.
14:44:30 <karsten> I mean, we could also schedule the copy for a time after which you have enough time to look into any issues.
14:44:38 <karsten> which might not be friday afternoon. ;)
14:44:44 <iwakeh> hehe
14:44:56 <karsten> what's your preference?
14:45:14 <iwakeh> hmm, hard to tell as I've not done it before ...
14:45:26 <iwakeh> better start from scratch and copy later.
14:45:31 <karsten> okay, I'd suggest to start from scratch. ok!
14:46:16 <iwakeh> all else by mail? and trac ticket?
14:46:51 <karsten> yep. we first need you in the onionoo group anyway.
14:47:00 <iwakeh> true.
14:47:11 <karsten> okay, I put a rough plan on the pad.
14:47:16 <karsten> next topic?
14:47:21 <iwakeh> fine.
14:47:29 <karsten> * CollecTor 1.3.0 (Webstats) release
14:47:44 <karsten> my inbox is my enemy. what tickets should I be looking at?
14:47:46 <iwakeh> actually, metrics-lib is the basis for this.
14:47:56 <karsten> ah, so metrics-lib 2.1.0 first?
14:47:57 <iwakeh> #22428
14:48:30 <iwakeh> its only a few changes here as most is in metrics-lib 2.1.0
14:48:55 <iwakeh> #22983
14:49:25 <iwakeh> this adds the new descriptor types, but changes very little on existing code.
14:49:49 <karsten> okay. what's the schedule here? end of july is so close..
14:49:55 <iwakeh> and also features tests for the new descriptors.
14:50:09 <iwakeh> true,
14:50:37 <karsten> though I could probably review two tickets and do a pre-release tomorrow.
14:50:41 <karsten> would that help?
14:50:51 <karsten> (for metrics-lib 2.1.0)
14:50:56 <iwakeh> fine.
14:51:02 <iwakeh> that is most of it.
14:51:15 <iwakeh> there are only few changes in collector then.
14:51:26 <iwakeh> it builds on the new functionality.
14:51:40 <karsten> and then we put out collector 1.3.0 on monday?
14:51:52 <iwakeh> fine.
14:52:34 <iwakeh> that way I
14:53:22 <iwakeh> could do some more testing, but it imports fine.
14:53:40 <karsten> okay.
14:53:48 <karsten> would be cool to get this done in july.
14:53:51 <karsten> as planned. :)
14:53:56 <iwakeh> yep :-)
14:54:05 <iwakeh> could you also
14:54:37 <iwakeh> review this #22912
14:54:59 <karsten> on the list.
14:54:59 <iwakeh> I added it to 2.1.0 as it will be needed for
14:55:09 <iwakeh> ah, fine :-)
14:55:21 <iwakeh> just two tickets for 2.1.0
14:55:28 <iwakeh> and one for collector 1.3.0
14:55:34 <karsten> ok.
14:55:39 <iwakeh> but they are big enough.
14:56:08 <karsten> okay, sounds like we have a plan.
14:56:14 <karsten> next topic?
14:56:16 <iwakeh> yep.
14:56:20 <karsten> * sanitized bridge descriptor spec
14:56:29 <karsten> regarding how to proceed.
14:56:29 <iwakeh> I noticed your comment.
14:56:40 <karsten> I though it's easier to discuss that here.
14:56:47 <iwakeh> yes, true
14:56:49 <karsten> not the details about the spec, but where to put it.
14:57:23 <karsten> we could put it on the third level, as in: Home >> Sources >> Tor Bridge Descriptors.
14:57:30 <karsten> which would be in parallel to collector.
14:57:45 <karsten> we don't really have a fourth level.
14:57:54 <iwakeh> you mean in section
14:57:56 <iwakeh> https://metrics.torproject.org/sources.html#main
14:58:14 <iwakeh> as Specifications
14:58:52 <karsten> to start with, yes.
14:59:05 <iwakeh> well, then we have a place.
14:59:18 <karsten> and the collector page would link to that.
14:59:39 <karsten> we'd probably take out most of the details about tor bridge descriptors from the collector page,
14:59:51 <karsten> because we should avoid keeping two copies of those details.
14:59:59 <iwakeh> un-clutter is always good.
15:00:15 <karsten> okay. :)
15:01:24 <karsten> there, plan added to the pad.
15:01:32 <karsten> the remaining parts can stay in trac.
15:01:41 <iwakeh> ok
15:01:43 <karsten> ah, one more topic:
15:01:53 <karsten> * MOSS final report
15:02:03 <karsten> you raised a good point there regarding deliverable 4.
15:02:13 <karsten> there's now an edited version that needs approval.
15:02:18 <karsten> from you and/or isabela.
15:02:25 <iwakeh> yes, I have that on my list.
15:02:32 <karsten> want to take another look and.... perfect!
15:03:14 <karsten> alright.
15:03:23 <karsten> nothing in my inbox stands out as super urgent.
15:03:30 <iwakeh> fine :-)
15:04:03 <karsten> okay, cool, we made good plans. let's execute them! :)
15:04:13 <karsten> thanks and bye!
15:04:14 <iwakeh> back to work!
15:04:16 <karsten> heh
15:04:25 <iwakeh> thanks, and bye, bye.
15:04:30 <karsten> #endmeeting