#metabrainz

in our case, and because the service is rather critical, I would use Always

rebooting mingus

... there is a genre called fungi?

wat

duh.

shroom fiends have their own genres, surely.

jajajajaja

zas: if you specify on-success, the process will only be restarted if it exited with a zero exit code

the consul service should only spontaneously die if it experiences an fatal error and will return a non-zero exit code in that case

so always looks better

i've figured out the reason why the services aren't starting on boot, i didn't add an [Install] section to the unit file

ah ;)

mayhem: should meb.o website be temporarily moved for rebooting trille?

the downtime should be less than 2 minutes, yes?

It's also running BrainzBot and MBSpotify.

Yes.

as long as we don't do it near the top of the hour, just a reboot is fine for 2 minutes.

do it at 30 minutes after the hour and tweet beforehand and it should be fine.

zas: I stopped cron jobs on trille, not sure about the two above ^

mayhem: okay, better start a new instance on another server then, only the redis instance will be missing.

new instance of what?

meb.o website container, I will use rakim

can't do that without moving all the large volumes.

which takes eons and loads of work to accomplish.

ok, I thought that it supported running from multiple containers like others.

sadly, no. this is the weakest part of our docker setup IMHO.

zas: Ok for you rebooting trille at 10:30 UTC (in 40min from now)? If so I will tweet about MeB.o maintenance.

I restarted MB cron in the meantime.

the new tagging feature on listenbrainz, will the tags that the user sets affect the track, the release, or the release group?

are the ackson5 servers at Hetzner or no?

jackson5*

yvanzo: yes (about trille)

zas: thanks, announced.

[metabrainz.org] 14yvanzo opened pull request #414 (03master…archive-annual-reports): Add archive notice to annual reports https://github.com/metabrainz/metabrainz.org/pu...

tandy1000: Last I heard, track/song/recording, to begin with anyway

okay, interesting, it would be useful to be able to configure that during submission, I'd mostly be interested in tagging release groups

yvanzo, zas: well, re wikidata-bot it might leave one edit half-done but tbh I don't think it's a big issue

reosarevok: is there a way to stop it normally?

(it's on zappa)

tandy1000: The UI for tagging in LB is tricky, monkey and I had quite a few runs at it. I think once the feature is in for tracks, then please bug us about making it better, especially if you can think of ways it could work UI-wise

yvanzo: not really, but most of the time it's not doing stuff, let me check

willdo! thank you :)

MB’s easy, we would just put in more buttons, no need to worry about crowding mum and dads phone display!

honestly if the MB UI / UX got better I'd be happy just living there haha

yvanzo: seemed idle right now, stopped it

(just with docker stop)

thanks

and by that I mean if MB had LBs new discovery features

tandy1000: Have you tried the Picard tag submitter? It’s magic (assuming your files have genre tags you’d like to add)

*Picard tag submitter plugin

no I haven't, i use beets

I need would something that grabs tags from bandcamp, rateyourmusic, lastfm

zas: zappa is ready to be rebooted (all the running containers have redundancy on other servers).

then i could submit those back >:)

picard vs. beets, the new vim vs. emacs

Please do restart the wikidata-bot container after reboot if it's not automatic

(I guess it might be tho)

Yes, grabbing tags from elsewhere would be very useful. A user a while back scraped all the Bandcamp tags (where there were links) and that must be a solid chunk of MB’s total tags tbh

there was some sort of bot adding tags from bandcamp to releases, I think alastairp might have been involved?

picard and beets don't have the same purpose, I use both, there is no "vs."

Yeah that’s the one. I think the user did most of it, with some help from alastairp?

yvanzo: that was very much tongue in cheek

I thought the purpose was to tag your files with MB data in both cases? But I haven't tried beets, so

yvanzo: ok, rebooting zappa

aerozol: yes, mr clon, that was a great thread

I’d like to have a cool UI where you can see tags from all levels of a entity - e.g. artist > release group > release > recording > work, no matter what entity you are on. But still be able to differentiate them (so you know where they’re from, e.g. if a recording is out of the artists usual genre)

Now THAT is a UI challenge. That I am not taking on today :D

If beets adds MBIDs the Picard tag submitter could still be useful

I could probably do something similar for rym and lfm, but IIRc there are some copyright issues?

aerozol: oh it does, but I don't have anything to submit from my metadata yet

Gotcha!

I have heard RYM is extremely un-keen on data scraping. I don’t know if LFM would care, but their tags can be very jumbled

yeah, i like musicbrainz cause only certain tags become genres

yvanzo: zappa is back

reosarevok: picard only has a GUI and is more about tagging album files, beets only has a CLI and is more about managing a whole library with its own local db.

Oh, ok. I didn't realize the local db part :)

reosarevok: has been doing magic with genres on MB for a while now, exciting stuff in the pipeline too (aliases!)

reosarevok, zas: thanks, restarted wikidata bot on zappa

Thnaks

thanks too

zas: ping trille

tandy1000: I don’t know what the official MeB line on that would be, but I guess we can’t stop you from submitting tags? Vaguely related, I have started adding links from MB release groups to RYM entities just lately, because I got a tool that made it easier

re tags ,that was. yea mr. clon! i loved that. i wonder where they went? havent seen thme in a while

oooh a lfm and rym genre tag bot woudl be ++ exccelent

as for copyright. those are user-submitted so lsf doesnt own shit

yvanzo: on it

scraping no. but maybe rym has an api?

Dug this up on the Discord: “RYM doesn’t only “not have an API”, it’s also extremely anal about scraping”

hmm.. when it comes to genres it's difficult with lfm, because they had only folksonomy tagging, same as us. but they nver "elevated" a set for genre i think

so you cna find old tags with "mr ownership deal" and thats my own tag for things iwned

From the users there that are familiar with RYM, this came up there… checks notes… last December :)

so not an api but aas

on the flip side i want desperatly to have imported a tagradio from there as a palylist inot lb

With lfm my concern would be where artists with the same name are mixed, but maybe there’s a way around that

oh yea. heh that'd be a pita

i thikn maybe jsut skip artists and albums alltogether and go with songs only

excet that they had no live vs remix vs alt vs original studio unless for osme reason it was named differntly.. very. not clean data

but i guess yo ucould do a run, and only use tags with x uses

to prevent "one time" ones. and also filter out "owned/seen live/etc"

yvanzo: trille is back

trille trall

Also from Discord: “My current setup was pulling data from last.fm for genres but it’s quite terrible even after applying an RYM genre whitelist due to no support for proper artist disambiguation”

does discord have chatlogs? and if ni, how cna we make that happen

zas: tweeted; looking into serge atm.

rebooting moroder

anyway noone is answering re the jackson5 servers so I can't; 1) add them to https://wiki.musicbrainz.org/Hosting/Server_names, 2) make a (new?) sectino for them, or 3) do anything else. ping me if there is any other thing

CatQuest: sorry, yes they are at Hetzner

what sort of thing are they? a separate section ?

also is there a way of forcing from your end a "stay logged in" session or an user on tickets?

can't select "remember my user" here becase i have to tunr js off in order ot actually log in

CatQuest: no separate section needed, they are servers like others, just acting together (spark cluster)

alrighto

may you give me the 5 names?

and who suggested?

servers still needing a reboot: aretha clash flobot floyd gaga kiss patton paco pink quest sentry serge

better have bitmap around for aretha/pink/floyd

zas (cc bitmap): serge is ready to reboot (I just stopped caa-indexer the same way we did for the schema change).

yvanzo: ok, I'll proceed then

yvanzo: paco should be possible too, it runs jenkins though

bitmap: I stopped cron sv for musicbrainz-json-dump container on aretha to make sure it wouldn't try to reach trille while it was rebooting. I started cron sv now. Just to let you know why it missed one hourly job.

yvanzo: serge is back

zas: yes for paco

yvanzo or zas, sorry ot interrupt, but may i have the name s of the spakr cluster servers? i will add thme to the server page wiki

CatQuest: jackie jermaine marlon michael tito

CatQuest: sorry but busy, I guess these are just the first names

🙇 thank you

3) is wrong, you can do anything else

yvanzo: ok to reboot paco?

zas: always

it as mostly to jsut have a 3'rd point :D

rebooting paco

here. https://wiki.musicbrainz.org/Hosting/Server_nam... they are there for now atleast

restored caa-indexer on serge

CatQuest: thanks

yvanzo: I *did* look at the mb page but alas, in true mb fashion, more than 6 names where linked as ember :D

member*

atj: reminder: we need to put docker packages on hold in Ansible

paco's back

what about flobot? It is AB

just reboot and not worry about it.

floboot

mayhem: ok, rebooting

zas: for patton, all MB cron jobs have to be turned off, can do next?

reosarevok: are we releasing today or tomorrow?

yvanzo: ok, let's go for patton

reosarevok: about to go to bed, but I wonder if re. auto-editor stuff it would be helpful to put a process into the wiki page for the community manager to follow. Even with email text in the wiki, or linking to a gdoc. Text like: “Congratulations! You are now an auto editor - here are all the things you need to know…”. If that sounds useful let me know and I’ll whip something up after I see your reply tomorrow

yvanzo: there are redis instances on patton

zas: that's why