zas: alastairp : did we chat more about moving jenkins from williams. I also want to deploy a typesense instance and using williams without jenkins might be a good fit.
ah, ok. when that is done, I'll start typesense on williams
alastairp
what's your thoughts on the cost of the VM? is there a limit we shouldn't go over? is it worth spending admin time trying to configure it to shut down when not in use?
(were you even thinking of using a vm? It was a suggestion that zas had while you were on the way to the office yesterday)
ruaok
I wasn't thinking VM, but that doesn't rule it out.
I think many of our bare metal machines are running below reasonable utilization and we should first look there.
alastairp
sure thing, let's go back and look at those when zas is available
v6lur joined the channel
Gazooo79494 has quit
Gazooo79494 joined the channel
Darkloke has quit
zas
alastairp, ruaok: I was checking VM prices, but we still have machines with enough cpu/ram/disk to host jenkins, for example cage, which is not overused atm
ruaok
then lets use cage.
zas
alastairp: ^^
alastairp
do we see any issues with running jenkins alongside mb website?
ruaok
I have no idea -- do you?
alastairp
bitmap mentioned yesterday that sometimes MBS selinium tests fail when json dumps are being generated, but that's the opposite - server load causing problems with tests
let me compare load stats on williams and cage
zas
well, I guess high cpu usage can slow down things
but if it becomes an issue we'll adapt
ruaok
if jenkins causes web site degration, it will show up on the graphs and zas will be on you like... a sysadmin on high latency...
alastairp
yeah, and it appears that there's no degradation by jenkins, it's the other way around
OK, will do the migration
zas: can you check if I have an account on cage?
zas
now you haver
alastairp
I'm in. thanks
sumedh has quit
_lucifer
alastairp: ruaok: Mr_Monkey mentioned in his review that you have discussed how to send updates automatically when the playlist is updated through the api. Can you fill me in as well :) ?
(in context of the lb websockets feature)
I had thought about it and am thinking of doing it using a rabbitmq queue. Wanted to know if you had the same approach or some other in mind?
Mr_Monkey
I'll add some more context: PR #1212 implements playlist updates through websockets, but currently requires the front-end to send a message announcing the playlist has changed (i.e user 1 modifies a collab playlist, once that succeeds the front-end page sends a WS message, then WS server broadcasts updated playlist to other users).
_lucifer: the whole thing depends on alastairp and I writing another "service layer" that emits the right sort of messages via web sockets based on the API calls that are being made. hopefully alastairp and I can work on that, (at least define it) this friday, if we can meet.
Chinmay3199 joined the channel
adhi001 joined the channel
sumedh joined the channel
reosarevok grumbles at https://tickets.metabrainz.org/browse/MBS-11291
reosarevok
yvanzo: the report should have run with the new code already today, right?
BrainzGit
[listenbrainz-server] alastair opened pull request #1231 (master…mock-playlist-metadata-lookup): Mock call to fetch_playlist_recording_metadata when getting playlists https://github.com/metabrainz/listenbrainz-serv...
[listenbrainz-server] MonkeyDo merged pull request #1228 (master…listening-activity-test-bug): LB-781: Fix failure in UserListeningActivity.test.tsx due to new year https://github.com/metabrainz/listenbrainz-serv...
alastairp
I think it's merge day in LB today. will try and get through most of the open ones
ruaok: if you have time for reviews today it'd be great to get your feedback on the 5 recent PRs that I opened on LB. they're all equally important, but there's no rush to get them all done today
_lucifer
ruaok: sounds great, thanks for the info!
ruaok
alastairp: I'm hoping to get to that later today. I've been reviewing the smaller ones already.
sumedh has quit
alastairp
thanks
yvanzo
reosarevok: sorry, I forgot to update cron container, that's why.
alastairp
I'll run through the depndabot ones now and then take a look at some of the older unmerged ones
yvanzo
reosarevok: updated
alastairp
ruaok: I'm still confused as to why we get this sporadic integration test failure, "psycopg2.errors.ObjectInUse: source database "template1" is being accessed by other users. DETAIL: There is 1 other session using the database."
alastairp: I dont think its the consul bit in this case, since jenkins is unlikely to use that. however, the create_db.sql for timescale has a reference to template1. maybe that is it?
alastairp
ruaok: agreed that it's not consul config. the timescale part is "create database using template1", which as far as I can tell shouldn't be a problem; it's just making the defaults explicit
I would prefer to either add a separate ignore/config file for eslint, or see what items we were actually ignoring and explicitly ignore them on the commandline
d4rkie joined the channel
D4RK-PH0_ has quit
Mr_Monkey: got some minutes to help me debug this?
alastairp: About the .gitignore file in LB, first of all if we want to ignore some files we can use the existing `.eslintrc.js` file (`ignorePatterns` option : https://eslint.org/docs/user-guide/configuring#... ).
I'm not sure exactly why the gitignore file was passed in the `format` command; the only path in there that could concern eslint is `node_modules`, which is ignored by default by ESLint (just checked the docs).