#metabrainz

monkey: thx for the suggestion, it helps a lot! another thought: desc is default for confidence, asc for other types. simplest way is to make asc default for confidence, but i'm not sure if that's what we want for users?

zas: Good morning. Just as information, I'm working on the tests for the collection class. Just need to pause because of work, will submit a PR later today

outsidecontext: ok, np :)

Hey, hi akshaaatt (IRC) ! Any update on my PR?

zas: just wanted to be sure we don't work both on it in parallel :)

ericd (IRC): Personally I think desc should be the default for dates and confidence, and asc the default for alphabetical sorting. I think asc as the default for confidence is not what users want. So I guess when the sorting mode changes you can set the asc/desc accordingly, probably overwriting the user's previous choice. Either that, or have a separate hasSelectedSortOrder boolean (or something like that) stored in the state to decide

whether to set the sort order when changing mode.

Does that make sense?

<07munishk> Hi @mayhem, I am working on the PR. It had worked earlier for lucifer as is so I am figuring out it maybe issue with packages

ok, great.

<07munishk> Trying running it on different environments

do the tests pass locally for you? they dont for me.

<07munishk> Not for me

<07munishk> But had passed for lucifer

well, work to make them pass for yourself first, then see about getting them to pass on github

<07munishk> Yep, working on it

monkey[m]: Are you using test.LB?

not i

Ansh negative, working locally on your PR review comments, feel free to use it !

munishk: i took a look at the PR, i think you missed my message on the changes needed to the url in the test last time. i commented it on the pr, that should get it working.

<07munishk> Thank you @lucifer

Hi pranav[m] ! Just reviewed your PR and I have left some comments

Went thru the review akshaaatt (IRC) will make the reqd changes by tonight..

Thanks!

[metabrainz.org] 14mayhem opened pull request #461 (03master…fix-invoice-number-assignment): Fix invoice numbers https://github.com/metabrainz/metabrainz.org/pu...

[metabrainz.org] 14mayhem merged pull request #461 (03master…fix-invoice-number-assignment): Fix invoice numbers https://github.com/metabrainz/metabrainz.org/pu...

hey everyone!!

any tickets i can look into?

LB-1040, LB-1380 look interesting, unless someone is working on those already

LB-1040: Last/libre importer improvements https://tickets.metabrainz.org/browse/LB-1040

LB-1380: add alt text support to the Art Creator https://tickets.metabrainz.org/browse/LB-1380

lucifer: https://blog.metabrainz.org/2024/04/05/off-topi...

🥳

yvanzo: https://www.irccloud.com/pastebin/Ntkc1wmE/

atj: great, did you have to set the ranges manually?

no, just specified shards and number of replicas

we could probably the set number of replicas to the number of cluster nodes for the smaller cores

I don’t really understand the ranges (such as `80000000-7fffffff`) but it probably doesn’t matter.

the number of replicas should probably be at least 3 rather than 2 in general.

found advice back: https://stackoverflow.com/a/63976603

yeah, I'll change it next week

I’m not sure that we need shards for anything but recordings

but if you think that we should have at least 2 shards for any collection with a dump > 1G, maybe works should be sharded too.

I'm not sure to be honest, it's total guess work at this point

same for me

I think I was looking at > 5G?

It's latency vs. parallelism from what I understand.

I looked at the dump size, not at the collection in solr data.

the number of entries is probably more pertaining.

I was looking at the index size in the Grafana dashboard

yeah, you're probably right

Last thing I would look into maybe is whether it is worth having 5 rather than 4 shards for recordings, given that we have 5 nodes only, if one node is down, there will be one node leading 2 shards.

do you mean 4 shards rather than 5?

yes

That is 40% of recordings instead of 25% if leading only 1 from 4 shards.

If we have only 4 shards, it might be more resilient.

well with 3 replicas every node will have several shards?

yes but not as leader

I assume that leader have more workload, but that is a total guess

yeah, I don't know how the queries are distributed

[musicbrainz-server] 14reosarevok opened pull request #3226 (03master…flow-component-syntax-reports): Migrate React report components to Flow component syntax https://github.com/metabrainz/musicbrainz-serve...