#metabrainz

mayhem: ignoring the Russian talking about some account in dig.tv (which does not even exist)

aerozol: if you make me the svg and png icons I can implement the pencil for testing

hey guys, I have a question. I'm working on a bot that indexes Genius and links Genius artists to MBZ artists. Is it really not possible to update artist relationships through the API?

vfosnar: no, you cannot add artist rels via the API. you need to use a library like mechanize, have a look at https://github.com/metabrainz/musicbrainz-bot

atj: the keydb issue happened again, reported as https://github.com/Snapchat/KeyDB/issues/675

mayhem lucifer Hi! Apologies for the delay. Got caught up in a lot of (inescapable (-_-;)) things...Coming to the point. First, I integrated lbserver with ClickHouse. It's able to consume messages from rabbitmq (those which are triggered by submitting listens). Msgs are generated in incoming queue, send to timescale, timescale deduplicates them and publishes to unique and clickhouse consumes those msgs using a bridge to the unique. The

msgs are stored in the required table while preserving the structure of the JSON messages.

Second, to test the querying speed, I made another db inside clickhouse and imported all the incremental dumps (from 15-05-23 to 15-06-23) in a table inside the db. I executed some queries and got some responses. I cannot tell whether the responses were fast or slow since my machine froze for the major duration of the execution. Maybe accurate testing can only be done with better hardware.

TL;DR clickhouse is able to ingest submitted listens via rabbitmq, and speed testing requires better hardware.

zas: probably worth strace'ing the process if it happens again

dravadhis: hi! sounds interesting -- and yes I bet ram is they key here. Do you have some code lucifer and I can look at and see how we can test this better?

We can use a structured approach and define the sub-columns (track_metadata.*) beforehand but then we'll have to know upfront all the possible fields and their types.

Actually this is cleaner

atj: I did (an strace), saved as strace.out on rex /root, I'll send it to them once I have time

ftr about reddit: apparently not even if you are a memeber cna you access "private" ones. i'mm a member of the lupinthe3rd sub and i can't access it becasue "not an approved member" apparently. evne if i replied to many posts and had discussions :/

so i dunno how it even works

<atj> "Vojtěch Fošnár ⚡️: no, you..." <- ah, thanks

zas: https://github.com/metabrainz/metabrainz-ansibl...

atj: great, work in progress of course, but I think that's the right approach at this stage

bitmap: do I understand from https://github.com/metabrainz/musicbrainz-serve... that the code is now running in the sitemaps container and sitemaps work again?

reosarevok: re: the pending edits icon, will you also be changing the icon used for pending artist edits? e.g. https://musicbrainz.org/edit/100664442

Hmm. I wasn't, but mostly because I forgot about it tbh

I can't see a reason why we shouldn't

bitmap, aerozol ^ ?

unless the "info" icon is also used for reasons other than pending edits I don't see why it can't be consistent

Yeah, agreed - I don't think there's any reason other than "this is older than that icon" :)

My only doubt is whether there can be a point where the same icon appears for entity edits *and* AC edits side by side

In which case we could have a third icon in the same vein but with like an artist icon and the clock :D

But hopefully not

i was wondering if there was a reason it has a proper tooltip

arsh: hi! I was hoping to see the work you've done so far. do you have any thing for me to look at?

atj: the new icons will *also* have a tooltop

*tooltip

To point to the actual edits, which is useful

ooh, shiny

Hi mayhem: Sorry I forgot to post the link that day due to the endpoint not working.

I believe the link is too long please let me know if you can access it or I can try something to shorten it

its working, thanks

i see the graph, pretty simple so far. is there anything I can do with it yet?

Did you try clicking any of the nodes

clearly not. ;)

It should present a new graph with that artist at the center

it does -- its fun already. :)

should Inandout be merged with I.A.O. then?

arsh: any way I can give it a seed artist?

I am trying to make that work currently

but I can add from my end

Do you have any artist in mind

8f6bd1e4-fbe1-4f50-aa9b-94c450ec0f11 please

Just a moment

For some reason its creating a new branch and not updating the main

So if you can please go to this link in the meantime

there, what should I do?

Ok I see there should be an option to change branches

Near the middle on the top

yes. which branch?

tender-keller

it not letting me access the page anymore.

Unable to access this workspace

Ok just a moment let me see whats wrong

Can you please try this link

For some reason codesandbox was not syncing with my github. Sorry about that

ok, cool that works.

it would be really nice to be able to visually show the strength of the relationship between artists.

right now they all appear more or less equal.

Sure

I believe we can make the first few artists appear closer

Should I also limit the number of artists as well because 100 seems a little overwhelming and slows down the process

agreed. lets start with 15-25 and see how that looks.

we may need to normalize the scale of the artists too.

Sure thing I will limit the artists

I am sorry if can explain what scale means

Does scale mean the number of similar artists displayed for an artist?

take a look at the score field for this artist.

zas: reosarevok I am getting read timeout on search on mb.org!

sporadic

So the score field should be normalized across artists?

arsh: given these two artists (portishead, blümchen) they have different ranges of the score field.

higher score means more similar.

arsh: yes, I think that may be needed for this to look right.

Ok I understand

👍

Basically the strength should be based on the scores

Sure I will start and work on that

Do you have any other suggestions in mind

mayhem: search timeouts out sporadically, if that's what you mean

It is known

Not sure exactly why tho

reosarevok: ok.

arsh: another suggestion is to have fewer bubble and to put the artist name inside the bubble.

having to mouse over a bubble to find out what artist it is, makes this much less compelling.

we should have fewer bubbles, but with text. perhaps with the use of ellipsis ...

Sure I will work on this feedback and keep you updated

Thanks for your help

np

mayhem: we had high load on certain solr nodes again, sir search.retry queue isn't emptied

ok

just making sure you knew.

solr response time is very bad atm, but I have no idea why

I'm restarting each solr instance in the cluster, one by one

solr issues seem to be fixed now, there were a high number of threads and tcp open connections on half of the nodes, apparently stuck

bitmap: perl reminder

I know %opts will expand to take any comma-separated things after the expected params, but is (stuff) equivalent to just stuff?

yeah

(override_privs => 1), foo => 2 will even be merged into the same hash

Sneaky

bitmap: updated https://github.com/metabrainz/musicbrainz-serve... then

thanks!

I just left a comment on the bot PR at https://github.com/metabrainz/musicbrainz-bot/p...

[musicbrainz-server] 14reosarevok opened pull request #2969 (03master…block-notes-on-add_note): Block notes on add note https://github.com/metabrainz/musicbrainz-serve...

Dropped the last commit from https://github.com/metabrainz/musicbrainz-serve... and added it separately ^

do you want me to focus on reviewing these first?

Pratha-Fish: are you available tomorrow at 15 UTC or so?

bitmap: well, the block notes thing is not so important

https://github.com/metabrainz/musicbrainz-serve... would be good tho

And I guess https://github.com/metabrainz/musicbrainz-serve... too