#metabrainz

mo’’in’

[musicbrainz-server] reosarevok merged pull request #1069 (master…MBS-10160): MBS-10160: Display username on sidebar for youtube/c/ links https://github.com/metabrainz/musicbrainz-serve...

MBS-10160: YouTube channel link that uses the https://www.youtube.com/c/username format won't have the username shown in the sidebar https://tickets.metabrainz.org/browse/MBS-10160

https://listenbrainz.org/user/Freso doesn’t have any listens from yesterday or today. Is something in the pipeline stuck?

hi gr0uch0mars

hi amCap1712

I saw that you opened a PR for recording details page

yup

looking forward to your review

Sigh.

yvanzo: around?

moooooin

reosarevok: yup!

amCap1712: I found a bug when navigating to a Recording from a Release. I put the error in the PR

Do you know what do I have to put in the line so it will read it properly?

Ok I'll check it out gr0uch0mars

thanks!!!

gr0uch0mars: does searching a release and clicking a recording work?

let me check

nope, same error in console

`… mbentity.Recording.getTitle()' on a null object reference`

Ok thanks found the error I suppose then, I must have forgotten to send the mbid of recording with the intent when clicking a recording in a release

Will fix it now

ok , let me know to re-review the PR

reosarevok: use utf8;

yvanzo: oh duh, thans

*thanks

Moin !

reosarevok: MBS-10108 is probably a style issue, do you want it to be moved under STYLE?

MBS-10108: twitter URL cleanup does not allow for links to individual tweets https://tickets.metabrainz.org/browse/MBS-10108

ruaok, iliekcomputers : I see the same as Freso: no LB listens since two days ago.

(or just green-lighting it?)

And look at the recent page :p

yeah. seems like the spotify import is stuck again. :(

I can try restarting the container.

ruaok: Not just Spotify. I have Simple Last.FM Scrobbler listens from yesterday that aren’t there either.

Mr_Monkey: And thanks for verifying. :)

yvanzo: oh, ffs. Tempted to say "Twitter videos don't count, fuck off with that noise" but I'm sure there's an artist out there somewhere who only puts music out as Twitter videos

No problem

confirmed spotify is not importing anything.

But if we allow this people will start adding tweet links to artists, because people will be people

yvanzo: can we whitelist this *only* for recordings?

sure!

And otherwise still clean it up automatically?

If that can be done, then style says do it

Yes, will only accept well-formated permalinks.

zas: https://stats.metabrainz.org/d/000000059/rabbit...

can you please set an alert on the listenbrainz queues growing?

if any of them grow past 1,000 it would be nice to get an alert.

400k is a problem. :)

yvanzo: I mean, if someone tries to post a link like https://twitter.com/mountain_goats/status/10623... for an artist, will it still automatically become https://twitter.com/mountain_goats ? :)

LB-450: Influx importer is stuck

iliekcomputers: ^^

everyone: when the LB listen stream seems to have stopped, check the status page: https://listenbrainz.org/current-status

and you'll get instant confirmation that something is busted.

reosarevok: we should probably just deny it as it may not link to the appropriate entity.

Eh, I guess

ruaok: moin!

moin pristine__!

ruaok: I was thinking to include cluster info in html files. People can help us that way, I guess.

[musicbrainz-server] yvanzo opened pull request #1080 (master…mbs-10108-tweet-rec): MBS-10108: https://github.com/metabrainz/musicbrainz-serve...

MBS-10108: twitter URL cleanup does not allow for links to individual tweets https://tickets.metabrainz.org/browse/MBS-10108

pristine__: what cluster info?

Number of nodes, executors, memory etc

makes sense!

And what are your thoughts on making the scripts independent?

depends on what you mean with independent...

I mean....Storing output of each script in hdfs.

Because presently we have an order,

First create_dataframes.py then train_models.py and at laat recommend .py

So for instance

If I want to just run recommend.py I need to wait till the other two scripts are done executing,

oh, yes. that is a must.

we should be able to load the model and run recommendations right away.

Otherwise we could just fetch necessary data from hdfs and run the script.

Apart from model, we need a few data frames too for generating recommendations.

These dataframes are the outcome of the first script run.

So yes, we can save every imp info/result.

great

Do we have something for today? I was just doing hit and trial to reduce time which can wait if we have something up.

how about some of the open old pending PRs?

Yeah. Good idea.

every time I go back to look at the old PRs they seem to be missing stuff. maybe take a break and get some of those done?

that way param can work on getting user stats working while you're on the recommendation stuff.

there are two

I will just go through them :)

yep. great. :)

zas: https://blog.metabrainz.org/2019/05/22/google-d...

spellew: looks right, To reproduce the issue in your local environment, please 1. update the docker-compose file image for `musicbrainz_db` to the new image (this is the one after schema change): `metabrainz/musicbrainz-test-database:beta`. 2. Remove `data/mbdata` directory and redo the steps for importing the sample dump (https://critiquebrainz.readthedocs.io/intro.htm...). It would be good if we make the changes

ahead of the coding period as we should be using the latest musicbrainz_db schema anyway. :)

yvanzo: did you forget the title on https://github.com/metabrainz/musicbrainz-serve... ? :/

Also, yvanzo: https://github.com/metabrainz/musicbrainz-serve... would be sweet to review when you have some time too :)

reosarevok: thanks, on it

bitmap, reosarevok: Module 'canvas' not found when running ./script/compile_resources.sh tests

Fail with any previous MBS release…

so the user was submitting track numbers as strings and then started sending them as ints

which led to the error.

i've added validation to the tracknumber field to keep it as always int for now, but not sure how to proceed for the current data, modifications to influx aren't easy.

iliekcomputers: hi, can we do an AB release today?

alastairp: i don't think this week will work. :( (monday?)

yvanzo: it's been like that for a while, yeah :/

if that's the earliest that you can, OK. We have a bit of a backlog for minimising frank load. I'm free now, so can get back on top of it

[listenbrainz-server] paramsingh opened pull request #574 (master…influx-bug): Add validation for tracknumber field in data https://github.com/metabrainz/listenbrainz-serv...

i can work the weekend, but i can't be fully working until friday.

ruaok: added the alert, but in fact there was one already, it's up since one month....

note: the lb proxy alert didn't happen since i changed sysctl values on lemmy

I'm busy most of the weekend, but ping me when you're availalbe and let's see if we can make everything work. I'll try and prepare everything

alastairp: sure, sorry about the unavailability.

np, we'll get to it when we can

reosarevok: I also used to run URLCleanup tests locally but it doesn’t work anymore since the move to ES6.

yvanzo: they work on static/scripts/tests/web.html

alastairp: brainzutils 1.9.0 is broken, may AB switch to 1.10.0 at least?