#metabrainz

yvanzo: Good morning

mo’’in’

Have you read my PR? :)

Yes and also tested it, it works nicely.

Thanks! Do you mean #2151 or #2114 ?

2151

I see, any suggestions on the code?

yyoung: yes, the "raw url" and "clean url" labels refer to the code and will probably not be explicit enough to editors.

Maybe change it to "URL" and "Cleaned up to" ?

That looks better indeed.

OK, and the variable names remain unchanged in the code, right?

yes

OK

Any other suggestion? :)

It lacks tests.

I know, just want to have your review first, to ensure I'm on the right way :)

If so I'll start working on the tests

yvanzo: Should I do something with the text overflow here? https://imgur.com/Kfyy7IS

Probably not, the full URL text must be shown for editors to visually check.

Alright.

When in doubt, ask bitmap and reosarevok too :)

Then should I disable text wrap?

OK :)

Do not hesitate to ask MB editors for feedback from UI changes, either on #musicbrainz channel or on community forums.

(They will be able to test it in beta, but if you want earlier feedback send them screenshots/screencasts.)

Roger.

yyoung: Anything else before I go?

yvanzo: Yes, look at this screenshot: https://imgur.com/mheBqCV

It occurs to me that the error targets still need to be enhanced

Yes

This is the default message for ERROR_TARGETS.URL

And the corresponding URLCleanup rule is indeed a regexp check for URL

But in this case it shouldn't target URL

I haven't figure out a reasonable solution though, so maybe deal with it later

yvanzo: That's all for now, thanks!

It should target the discography entry.

Yes but the problem is URLCleanup did not provide enough information

Right, the 'validate' function for 'amazon' should be enhanced.

Not only the URLCleanup stuff actually.

When it validates one relationship and not the other for the same URL, the target should probably be the second relationship.

Hmmm maybe we can check whether there're multiple relationships under this URL and react accordingly

Yes

My current mechanism is to display URL-targeted message under the input box

But the validation is carried out for each relationship

OK then, I'll try to fix it later

I'll start working on the tests, and ask editors for their opinions on the UI change the other day

yyoung: OK, keep up the good work! :)

zas: Some search index updates from Monday were missing, I resent them.

But a full rebuild of search indexes will probably be needed for earlier missing items.

If search.index queue is messing up again, just stop sir-prod on pink to stop it writing unneeded large logs.

zas: hi! ssh connections to bono are timing out. can you please help?

https://similarity.acousticbrainz.org/ is also hosted on bono and returns a connection timeout as well.

timeout for me too

it doesn't answer to ping either, I reset it

lucifer: server is back

zas: thanks! able to login now.

[bookbrainz-site] 14akashgp09 opened pull request #657 (03series-entity…master): Merge master into series-entity https://github.com/bookbrainz/bookbrainz-site/p...

alastairp: ^ trying to train dataset on bono raises the above exception

lucifer: yeah, it can be temperamental. how big is the dataset you were evaluating?

2 classes, ~150 recordings each.

that _should_ be enough to prevent the usual errors

at a guess from the error message, one of the preprocessing tries to delete the ['metadata'] block from the ll json, but maybe AB has filtered that out before?

maybe this failOnUnmatched could help?

if not, we could temporarily try and delete the metadata.* from here: https://github.com/MTG/gaia/blob/master/src/bin...

i see makes sense.

so passing failOnUnmatched=False here should be enough?

unsure - I don't know if our project template file allows this parameter to be changed

https://news.ycombinator.com/item?id=27672508 discussion about beets, picard gets mentioned a lot

ah ok, we can do it https://github.com/metabrainz/acousticbrainz-se... then

lucifer: yes, that's probably better. however - 1) double-check the git tag that we check out when compiling gaia, that feature may not be present in the version that we're using

and 2) this process did work, so I'm a bit interested in working out why we're running into problems with it - I suspect that there's a solution to this problem that means we don't need to make such a drastic change

[bookbrainz-site] 14akashgp09 opened pull request #658 (03series-entity…series-collection): feta(collection,series): Extend Collections to include Series Entity. https://github.com/bookbrainz/bookbrainz-site/p...

yeah, you're right, we're checking out 2.4.5. the failOnUnmatched is only present in latest master.

2 makes sense as well.

unfortunately we may have to dig into the gaia code, but I'm a bit surprised that it's buggy

which branch are you running this on?

this one https://github.com/metabrainz/acousticbrainz-se...

on the bright side, sklearn eval completed successfully

ok great, nice to confirm that the end-to-end works there

I wonder if there's some changed code that accidentally runs during the gaia flow that shouldn't

yeah, i would want to try out the dataset on AB prod and see if it works there.

should we do a release first?

yeah, sure!

nice! :D

any PRs ready to merge or should I create the release?

no, everything merged for now

👍

[acousticbrainz-server] release 03v-2021-06-30.0 has been published by 14github-actions[bot]: https://github.com/metabrainz/acousticbrainz-se...

loujine: thanks for sharing that link

updating AB prod.

outsidecontext: np

alastairp: AB didn't start as expected.

looking

i have reverted to previous image.

this was the log before i shut down container from today's image.

uwsgi didn't start up or exited with some errors proabably?

there was nothing after the enabling syslog message?

oh my bad 🤦

seems like consul was waiting for config items that don't exist?

yes right

this is not merged yet

sure

go ahead with that one

👍

up now.

nice

all AB containers updated.

did something related to AB volumes change recently?

we moved servers

I guess we didn't copy the contents/permission of that volume

but that should be in the docker-server-configs scripts?

what should be? the definition of the volume, yes

but given the problems we've had with permissions on dumps on LB, I'm not surprised that AB volumes don't start up properly either :)

ah right the definitions are there.

well, good thing boingo isn't decommissioned yet :)

so we can fix the definitions of the volumes in the Dockerfile, but we should also copy the contents of the volume from boingo

yeah. I'll do that now

we should set up backup to create backups from clash too

(and disable boingo)

+1

lol, suddenly my list for today got really big

lol XD

let me if I can help with some of it? :)

*let me know

thanks, but I think I'm good

lucifer: unrelated, I ran into https://serpy.readthedocs.io/en/latest/ yesterday

something to keep in mind in a possible future

yes indeed. might be useful to us.

[acousticbrainz-server] 14alastair closed pull request #366 (03master…similarity_paths): Similarity Path Hack https://github.com/metabrainz/acousticbrainz-se...

[listenbrainz-server] 14mayhem merged pull request #1527 (03master…add-MBID-to-pinned-recording): Add optional recordingMBID to pinned recordings https://github.com/metabrainz/listenbrainz-serv...

rdswift: should i merge 1536?

ah, sorry. ruaok ^