#metabrainz

working in a park though, ftw

it's much to cold to do that here :D

also public internet is shit :(

ruaok o/

Good to see you back :D

Ping me when you are free for a discussion about current solr status and how we are going to proceed

samj1912: :)

samj1912: might as well be now.

Cool

So we are upto feature parity with current search server

I also fixed all the bugs blocking the initial release

great news!

all?

The last replication bug has an open pr for review

Yes all

wow.

Over the weekend I worked with zas dockerizing the solr server

The solr part is done

wasn't that done already?

Not properly

The sir part is dockerized according to bitmap and so are the rest of the requirements

what other docker parts need work?

Once this latest PR is merged we can test the live update queueing process.

I made a doc

1 sec

15:11:24 <samj1912> https://docs.google.com/a/metabrainz.org/docume...

nice graph!

Everything there is dockerized. We only need to test it once if it's all working together in conjunction

great.

so when that last PR is merged, then setup the live replication stuff and monitor it.

once that seems to be behaving, point beta.mb.org to it.

Yes

Cool

then we'll write a blog post about a call to test.

Uhuh

we'll get a flood of bugs, for sure.

we'll have to be careful to curate them about parity and new features.

I don't want to go down the rabbit hole of new features right now.

I think we have reached parity

think being the operative keyword. :)

we probably missed something, but that is expected.

and if not, bonus!

then once we go live with the setup, you should take some time to work on picard.

I went through search server source code, but yes, testing will be nice, we can also get a measure of how well solr performs under load

but, the primary focus is search still. If bugs in the setup appear, that is top priority.

Okay

thank you again for getting a handle on all of this. I feel uplifted now. :)

⬆️ ⬆️ ⬆️ ⬆️

🙌

alastairp: ping

iliekcomputers: I'm going to nuke the alpha-importer bits. should be pretty easy.

Meanwhile, I wanted to have a discussion with you and bitmap about the existing search server schema

It's not really all that comprehensive

what do you mean by that?

Lots of things are missing or inconsistent as me and bitmap figured out. We fixed or not added whatever we found

Well for example the search server indexes only url artist rels

Nothing else

It had different relationships outputs depending on which type of relationship it was (missing attributes and so on)

And has a slightly different output than WS2

ruaok: I have it done almost, I'll send the PR in a half hour or so.

this is present in the old search server code and now also present in the SOLR stuff?

iliekcomputers: ahhh, ok. thanks. <3

Present in old server yes

Solr too (although I took bitmap 's advice to remove or add stuff which made sense)

But yes, solr is close current search server than WS2 in output

Since I coded it entirely replicating the search server Java source code

oy, this is tricky stuff.

yes, lets chat about this with bitmap.

Yes

It would make sense to add as much as we can that won't break other people's setup (like the rest of the URLs) but there might be stuff we can't fix :(

and that no one ever noticed. :)

Yeah, I though that's what the discogs importer for example used to try to match discogs labels and places and releases, but I guess not

ruaok the backlog where me and bitmap discussed this https://chatlogs.metabrainz.org/brainzbot/metab...

I guess we took so long to add it at first that everyone was using some other option by then :p

samj1912: you and me both.

if someone links me the backlog, there better be good reason to dive it. such a pain.

Isn't that URL a proper link?

(just copying the current URL isn't, but that should be, no?)

it is a proper link, but it doesn't display as one would expect.

[listenbrainz-server] mayhem closed pull request #270: LB-223: Add incremental dump table (master...dump/add-table) https://git.io/vd18e

[listenbrainz-server] paramsingh opened pull request #282: LB-236: Remove import from alpha feature (production...alpha-import/rm) https://git.io/vFtcP

[listenbrainz-server] mayhem closed pull request #282: LB-236: Remove import from alpha feature (production...alpha-import/rm) https://git.io/vFtcP

samj1912: How is that link not a proper link to the backlog?

Freso it does not scroll to the correct place. You have to find the correct message.

samj1912: Did for me. 🤷🏼

Firefox? :p

yes.

Chrome problems then :p

For me it scrolls to an arbitrary message later that day

I have to scroll up to find it

Even worse is going up through the backlog normally. It will load the backlog and then scroll me down to the bottom.

bitmap, Leftmost (and maybe chhavi_, samj1912, iliekcomputers, ferbncode-who's-not-around): Europe switched DST off last night, so tonight's meeting (and coming meetings) will be at 18:00 UTC rather than 17:00 UTC. AFAIK, the US doesn't switch until next weekend(?), and no clue about India. :x

Ouch :p

samj1912: Curious.

India does not follow dst

I thought they mightn't. :/

I don't think chhavi will join us that late

:( Oh well.

So meeting at 11:30 PM IST today

?

Yes

iliekcomputers: https://github.com/metabrainz/listenbrainz-serv...

Can we abolish it already

Ruaok: Whoops

iliekcomputers: the importer is still using the beta url. I guess that needs to change to a production URL now.

ruaok: doesn't it use api.listenbrainz.org?

We could change the name of the variable.

it does, but the config settings are now confusing.

yes.

ruaok: Ok, on it.

thx

make that fix on master, please.

Timezones are officially a pain

ruaok: why on master?

it impacts hosting changes.

simpler this way.

master still has alpha importer so it might break stuff.

Maybe we should merge production into master before doing it then?

yes