[00:13:42] <inflatador>	 latersville, see ya tomorrw
[00:53:58] <ebernhardson>	 heading out too
[10:19:37] <gehel>	 Do we have documentation for the format of the RDF Stream of the streaming updater ?
[10:20:36] <gehel>	 Wikimedia Enterprise is interested in Wikidata. Good for us!
[10:24:49] <volans>	 will they pay for it? :-P
[10:29:33] <zpapierski>	 we have schema, but since it's not yet public, we didn't really do anything else, unless dcausse did
[10:45:59] <dcausse>	 I started this https://gerrit.wikimedia.org/r/c/schemas/event/primary/+/594098 but not it's not particularly self-explanatory
[10:59:40] <dcausse>	 lunch
[11:02:29] <zpapierski>	 ah, so there isn't even a schema
[11:02:39] <zpapierski>	 but we do have a task for publicizing
[11:08:11] <zpapierski>	 T294133
[11:08:11] <stashbot>	 T294133: Expose rdf-streaming-updater.mutation content through EventStreams - https://phabricator.wikimedia.org/T294133
[11:08:44] <zpapierski>	 lunch
[12:18:38] <ejoseph>	 errand
[14:10:58] <inflatador>	 My internet is flaking out again
[14:11:04] <inflatador>	 I'm on hotspot, shouldn't effect work too much hopefully
[14:11:12] <inflatador>	 err...Affect that is
[14:13:34] <dcausse>	 o/
[14:15:05] <dcausse>	 on my side I finally seem to be elligible for fiber
[14:24:29] <inflatador>	 we had some winter storms last week, I think there may be some damaged lines around my neighborhood
[14:27:29] <zpapierski>	 o/
[14:27:40] <zpapierski>	 dcausse: so lucky, it will be years before I get mine
[14:27:57] <zpapierski>	 fortunately, it turned out I can actually get a semi-decent connection on GSM here
[14:28:21] <zpapierski>	 slower than I had in the previous apartment, but I think more stable
[14:28:46] <dcausse>	 it's what I have atm, 4g and a antenna on the roof and speed is quite descent, way faster than ADSL
[14:29:14] <dcausse>	 sole problem is busy hours 5-6pm connection becomes flaky
[14:29:15] <zpapierski>	 antenna is something I want to use, without it I get about 90-100Mbps/20-30MBps
[14:29:25] <zpapierski>	 yeah, same here 
[14:36:17] <inflatador>	 My ISP's customer service is getting worse too. Maybe it is time to switch over
[14:38:18] <zpapierski>	 I'm trying to build rdf and it complains that "address already in use", it's a start-proxy stage and the port is 8812. I have nothing there. Didn't have that issue before... dcausse - sounds familiar?
[14:39:39] <dcausse>	 zpapierski: this seems to be the integration tests, I thought we've got rid of this proxy
[14:40:02] <zpapierski>	 that's the patch with flink update
[14:40:10] <zpapierski>	 I'll rebase and see if helps
[15:23:15] <zpapierski>	 it did
[15:30:15] <ebernhardson>	 can someone create a ticket for https://gerrit.wikimedia.org/r/c/wikidata/query/gui/+/759775 ? For some reason i'm unreasonably annoyed and not writing nice things
[15:30:19] <ebernhardson>	 also, \o
[15:38:42] <dcausse>	 o/
[15:41:11] <dcausse>	 :/
[15:41:52] <ebernhardson>	 i should just write it...i suppose it's just a compounding of annoyance at things that don't work on wcqs, and how phabricator is a place tasks go to wait a month before doing anything
[15:44:06] <zpapierski>	 o/
[15:44:38] <zpapierski>	 nothing to be annoyed at here, but ticket to discuss feasability of query builder for WCQS along with disabling the link for now makes sense
[15:44:51] <zpapierski>	 want me to write it, or will you do that?
[15:46:23] <ebernhardson>	 zpapierski: please write it :) The thing is i guess i don't particularly care about if the query builder works or not, i care about if clicking around on commons-query.wikimedia.org makes sense. This is just another in a long line of things where wcqs is a secondary-thought
[15:46:47] <zpapierski>	 well, it sort of is :)
[15:47:05] <zpapierski>	 like, literally
[15:47:14] <ebernhardson>	 but that's the worst way to live your life
[15:47:30] <zpapierski>	 fortunately WCQS isn't sentient yet 
[15:47:44] <ebernhardson>	 :P
[15:54:22] <zpapierski>	 T301255
[15:54:22] <stashbot>	 T301255: Adapt QueryBuilder for Wikimedia Commons Query - https://phabricator.wikimedia.org/T301255
[15:56:13] <ebernhardson>	 I guess that ticket kinda still describes my problem, adapting query builder is a long task that requires lots of decisions. All i wanted was to make clicking around not confusing but that ticket suggests we will evaluate options, integrate, and some time after users are trained to not click that button it will be fixed :P
[15:57:05] * ebernhardson feels like what should have been a 1-line fix is instead a long drawn out process
[15:57:35] <ebernhardson>	 i guess i'm just grumpy this morning :(
[16:01:07] <dcausse>	 touching the wdqs UI is not a good way to start a morning :)
[16:12:11] <zpapierski>	 ebernhardson: I think we should definitely start with removing that link, whatever the decision will be in a long run
[16:18:03] <inflatador>	 Going back to primary Internet, wish me luck
[16:22:22] <inflatador>	 anddd, we're back
[16:53:59] <ebernhardson>	 dcausse: lgtm, ship now or later?
[16:54:29] <ebernhardson>	 i'll also be shipping the query_clicks -> airflow later today, hopefully at least
[16:55:21] <dcausse>	 ebernhardson: you can ship but not activate, I'll have to adjust the start dates
[16:55:44] <ebernhardson>	 dcausse: kk
[16:56:07] <dcausse>	 that means no more oozie jobs, cool :)
[16:56:52] <ebernhardson>	 stalled long enough, few years? I seem to have this bad habit of not finishing things...
[16:57:17] <ebernhardson>	 get there eventually :)
[16:57:19] <dcausse>	 I think it's a pretty common habit here :)
[16:57:29] <ryankemper>	 :P
[16:58:23] <ebernhardson>	 i guess i should be more impressed oozie kept running it for ~2 years with only 2 or 3 minor deployments to fix the namespace partition
[16:58:47] <ryankemper>	 gehel: inflatador and I brought the new eqiad masters into the cluster yesterday and booted the old ones [from master eligibility, not the cluster]. all that's left before actually decommissioning is to evict all of the old nodes from the cluster
[16:58:51] <ebernhardson>	 might be hard to use, but was reliable i suppose
[16:59:34] <ryankemper>	 gehel: I think the best way to do that would be to just ban all the hosts we're going to decom from the 3 eqiad elastic clusters all at once. sound reasonable? (will presumably take several hours before shard reshuffling completes and we can actually decom)
[17:00:26] <gehel>	 ryankemper: yep, sounds good. The rate limiting should protect the cluster.
[17:00:38] <ryankemper>	 ack
[17:00:43] * ebernhardson is more wondering about if we can change rate limiting after kicking out these old 1G instances ;)
[17:00:58] <gehel>	 dinner
[17:03:50] <inflatador>	 quick workout, back in ~30
[17:15:38] <ebernhardson>	 hmm, seems the entire output of a script we invoke from scap is turned into a single line, logged by scap on the deploy hosts, and triggers OSError: [Errno 90] Message too long 
[17:16:13] <ebernhardson>	 but it still "works" and claims success. hmm
[17:28:30] <inflatador>	 back
[17:29:38] <ryankemper>	 ebernhardson: yes we should be able to flip to 10G after...I'll need to do a scan to be sure though
[17:42:08] <dcausse>	 dinner
[18:27:49] <inflatador>	 ;unch, back in a few
[18:27:53] <inflatador>	 er.r...lunch that is
[18:57:12] <inflatador>	 back
[20:05:14] <ebernhardson>	 hmm, cirrus docs might be slightly outdated: NOTE: this list was written for 0.90 so it may not work well for 1.0.  It'll be revised when I have more experience with 1.0.
[20:20:22] <ebernhardson>	 also hmm, not getting a performer field from wcqs-external.sparql-query although it is submitting the correct schema version onw :S
[20:20:24] <ebernhardson>	 *now
[20:29:00] <ebernhardson>	 lunch, will poke later
[21:11:19] <ebernhardson>	 back
[21:18:36] <ryankemper>	 ebernhardson: you around to give some insight on elasticsearch shard scheduling behavior?
[21:18:45] <ebernhardson>	 ryankemper: i can guess :)
[21:19:42] <ryankemper>	 ebernhardson: http://meet.google.com/bxh-wxfp-wty