#metabrainz

listenbrainz is still up!

zas: entry for kiss added to metabrainz.org domain records.

ruaok: ok, thanks. Server not visible yet in hetzner interface

zas: we got pings!

yup

I can't login yet. I wonder if they are rescue mode

looks like it is rebooting now.

its back, but I still can't login.

lucifer: can you double-check kiss.sh versions on https://github.com/metabrainz/docker-server-con... ?

zas: root passwd in mail

yup

ok, go for it, zas. let us know when its done.

I'll stay out of your way.

fuck, you're right we didn't order the extra nic

alastairp: reviewed.

ruaok, we also need the policy on beta and test so let's remember to update those when kiss comes up.

k

[listenbrainz-server] 14alastair opened pull request #1487 (03master…dumps-no-mount-check): Remove check for mounted dumps location https://github.com/metabrainz/listenbrainz-serv...

zas: I've opened the request to have the second nic installed.

lucifer: 27.0 is the current latest?

yes

ruaok: ooops ;)

we'll need to shut it down again, once they come back to me on that.

oh wait. I thought I set that to 27.0 werid

ohh, I set cron to 27.0 and skipped web. that's it

yup

we need to deploy 1487 anyway

right

so let me just update all containers to 27.1 now

i am hoping for blitz build because no dependency changed 🤞

[listenbrainz-server] 14MonkeyDo merged pull request #1418 (03master…importer-progress-bar): Importer Progress Bar https://github.com/metabrainz/listenbrainz-serv...

Sneaking another PR in

[listenbrainz-server] 14alastair merged pull request #1487 (03master…dumps-no-mount-check): Remove check for mounted dumps location https://github.com/metabrainz/listenbrainz-serv...

lucifer: want to try and make the release?

alastairp: sure, i think we can get another two PRs in. 1479 and 1480, should we get those in first?

looking

Anyone want to volunteer as test subject?

what does it ban on? name or ip?

alastairp: Your account, for the purpose of this test.

🙋‍♂️

what's an account? nickserv?

nickserv it is :D

alastairp: Yeah. :)

Alright, looks like it’s working. Great. :)

Thanks lucifer!

lucifer: updated https://github.com/metabrainz/docker-server-con... to 27.1

I'm currently running tests locally to check https://github.com/metabrainz/listenbrainz-serv...

but looks ok

[listenbrainz-server] 14amCap1712 merged pull request #1480 (03master…do-not-sanitise): LB-767: stop 'sanitising' docker compose container names https://github.com/metabrainz/listenbrainz-serv...

alastairp: https://github.com/metabrainz/listenbrainz-serv... i agree we should be consistent. but can we consistently start using f-strings instead, % seems like python2 way of formatting?

in short, no.

this is what I was commenting in the previous PR, we should decide how we want to do logging. either we do their recommendation and don't do string interpolation and let the log function do it. in that case we have to use %, because that's all that's supported

otherwise we interpolate it ourselves and use .format or f-string

Updating beta.mb.o

as I mentioned, I don't really mind what way we do it, but we should be consistent

(the reason that you do the `debug("some % string % here", param1, param2)` is so that you _don't_ have to do the interpolation if your debug level is such that you're not going to output anything - in the bad old days when string interpolation was expensive this made your code run faster)

oh - maybe I'm wrong. just looking into this in more detail now. it seems that you might be able to use {} placeholders in logging functions

https://docs.python.org/3/howto/logging-cookboo... that's pretty scary

alastairp: sorry. i had confused your % formatting comment with something else. debug("Hello %s", " World!") is fine. i meant to use f-string in cases outside logging, at a few places we do stuff like "%s %s" % (a, b). i was wanting to suggest using f-strings there.

yes, right. there's _another_ question to open about how we do string interpolation in general :)

sorry, my comment on that PR was specifically about formatting in logging calls

ruaok: During this connection process ListenBrainz will request permissions to view your YouTube profile data. During the connection process ListenBrainz will request permissions to view your YouTube profile data.

fun.

lucifer: ^^ 🤦 fixed in master now.

just in time for the next release :D

sigh, hetzner.

[listenbrainz-server] 14amCap1712 merged pull request #1479 (03master…lower-log-level): Log recommendations without users to debug only https://github.com/metabrainz/listenbrainz-serv...

you didn't rub your tummy before spinning around waving the dead chicken in the air.

your request is DENIED.

[listenbrainz-server] release 03v-2021-05-27.1 has been published by 14github-actions[bot]: https://github.com/metabrainz/listenbrainz-serv...

yvanzo? Did you see the issues I mentioned yesterday?

hetzner> If you require this as well on the new one, you may please submit a separate support request.

what in the actual fuck is going on now?

built in ~3m45s, js was recompiled and that took almost ~1.5min so ~2min if only python code changes without dep changes.

well done lucifer, this looks like it'll work great for us

let's try and get an irc bot next that lets us make a release automatically :)

yup, that's up next :D

i'll wait for another couple of releases to be safe and then add the action to all other PythonBrainz

ruaok: where are we at?

30k items in listen queue, maybe we should start up the writer after all

waiting on hetzner.

ok, going for it

zas: pings!

the machine is up, eth1 not configured. I supposed that is expected.

are the nics behaving, zas?

yes, I'm on it, that will take some time, I need to update a bunch of things

alastairp: available for discussing on making request consumer a daemon

?

what's the request consumer?

(yes)

the LB entrypoint to spark cluster

the spark side that takes requests

which machine does it live on, and how do we currently run it?

with nohup and &

it lives on michael.metabrainz.org

nice

I'd be tempted to just make a systemd file

acc to my recently gained understanding of runit, we can just use runit like we do in other service

lucifer: that code looks legit. just copypasta that.

alastairp: systemd is for starting it, but the python program itself needs to daemonize.

systemd can handle processes that stay running in the foreground

how is the request consumer different from spark reader or timescale writer (except the docker part)?

lucifer: it doesn't run inside docker.

where we have runit that automatically starts and monitors services.

right makes sense.

runit expects that the thing that you start stays running in the foreground

other init tools can handle processes which daemonise

ruaok: mm, what consumes playing now messages?

I'm not sure we've been reporting them properly

makes sense. request consumer should be a background processs methinks

redis_listenstore

ultimately which goes into websockets

lucifer: agreed, it should be background,

see the first few paragraphs of this answer, lucifer: https://unix.stackexchange.com/a/308394

if you copy the code that I linked and use that for startup, you're sorted.

https://listenbrainz.org/user/mr_monkey <- no now playing

are we missing a websockets server?

quite possibly.

did you start it on boing?

i do not see any websockets srever on boingo

+o

should i start it?

I missed some of the conversation.. alastairp, do you have or recommend that mug?

lucifer: go ahead

up now

that was a miscommunication, ruaok thought that I started it, and I didn't know that it had to be started. no problem

quite a steep fall :D

that queue went down quite quiclly

alastairp: i read that answer so foreground processes are simple.

acc to systemd