#metabrainz

<12Aerozol> @pranav: for following v following, those questions are easy - just copy exactly what desktop LB does! We are just copying those components anyway

<12Aerozol> @pranav: oh, I didn't realize (or more likely forgot) that we already had the similar user app component! I would say no need to redesign it, but if we find it ruins your lovely design work on the rest of the user page then let's revisit it

<12Aerozol> @jigen: is this the issue you're encountering? https://tickets.metabrainz.org/browse/MBS-13538

MBS-13538: Removing all rows in the artist credit editor makes it disappear

<12Aerozol> Seemed pretty serious to me too, but I thought maybe it was just me

@zas Sorry - searching still dead.

<10volumetrique> Is the search cluster still down?

<10volumetrique> Doesn't work on my side

It should work, but the rate limit was low, so it fails very often. It was to let the cluster recover, I just restored the default limit. Let's hope it doesn't crash again

<10volumetrique> 🫡

<10volumetrique> Seems to work better now

<10volumetrique> At least direct search always work 😅

pranav: btw, I tagged you in a few replies above, in case it didn’t ping you

pranav: Also, I have updated the user page designs. I think these should be 100% signed off now - EXCEPT for playlists and created for you, I still need to know if you plan on doing these, and if so, to what level of detail (e.g. what parts of the desktop features you will or wont incorporate)

Link: https://www.figma.com/design/tFLTQiq7QlA1oGRTJV...

pranav: I have also finished a first pass of the artist page. It’s a bit of a beast! Do you think you could have a look at it monkey ? Keen for any feedback on anything that could be improved or won’t work: https://www.figma.com/design/tFLTQiq7QlA1oGRTJV...

If anyone’s curious, here’s a export of the artist page. The figma version comes with notes though. I’m still not toooo happy with the links section, but I think it’s better than it was

This makes me think that being able to minimise sections would still be pretty good pranav I’m probably over-engineering this, but this could be a good way to make it not accidentally accordion all the time/get annoying:

Very slight accordion arrow, which gets clearer + instructions to double tap, if tapped once.

That’s me for now. Night all :)

What's up with the LB stats being stuck on the 10th?

<yvanzo[m]> "The initial surge of index..." <- Any ideas why we don't see a similar peak in updates on the new Solr cluster? https://stats.metabrainz.org/goto/voZQm-USR?org...

nothing at midnight but then a peak between ~08:25 and 09:05 UTC

aerozol: YES that's it!

this honestly is kinda a vital bug .. higher crit than currentlythe ticket imho

uhm. is it really too much to ask that some kind of backup was made of my liked tracks before my profile was wiped? https://beta.listenbrainz.org/user/ApeKattQuest...

like is there a dump or osmething atleast where I can manually redo them?? seriously?'

also wam playlists? they're also completely gone! why?

couldn't you havetold me beforehand soi could have idk,exported or noted my things first :/

anyway, when will stats be recalculated?

also global stats for all time is broken

atj: It is likely that the bottleneck was the SolrCloud 7 instance which had to handle the search requests from production too. See https://stats.metabrainz.org/d/000000059/rabbit...

atj, zas: Also from SIR (debug) logs, it seems that committing Solr was the slowest step in indexing. ^

commits should be ignored

but perhaps I'm misunderstanding

atj: Not just committing, sending data to Solr as a whole is the slowest, compared to retrieving data from Postgres.

I can’t compare with the usual time to send data to Solr though.

yvanzo[m]: right, makes sense given the load averages on the nodes were probably sky high

(I haven't looked at the graphs yet)

s/is/was/

can you stop SIR sending commits?

they don't do anything in prod so are pointless

atj: Do you mean for SolrCloud 7 in production, SolrCloud 9 in beta, or both?

yvanzo[m]: https://github.com/metabrainz/mbsssss/blob/mast...

both AFAICS

that configuration tells Solr to return 200 to all update requests

word of the day is jugaad

might be useful for the summit!

in listenbrainz, when does stats update after an import?

for an user, but also global and world wide

Jigen (IRC): I saved a listenbrainz dump from June 1st in case it can be used to restore your loved tracks, but not sure it has any playlists

do you know how your account got deleted?

hey akshaaatt!, I am working on the UI bug, I have tried changing the search bar entirely, maybe the issue is in how we are calling it in the views, looking at it.

atj: Thank you. Correct that it should be disabled it in SIR too for clarity, but it won’t have any impact anyway.

Should only have an impact under high load when Solr is slow to respond to any requests

Removed committing from sir-prod already, restarting it too.

One newer feature of HAproxy is request prioritisation, I meant to try it on beta to prioritise requests from SIR

can't really think of a good way to test it though

We can test it in production for a short while if needed.

[18:52] bitmap[m] do you know how your account got deleted?

absolutely 0 idea. i checked it yesterday for some reason and it was totlaly empty

i assumed it was something that happened due to the weird user updating bug casued by that (bot?)Account with the kanji name

so it wasn't actually something you guys did/something that happened due to the bug :O?

ohh!

i assumed it was osme ting you guys did to fix it :o sorry

no idea what happened srsly

but if you could give me back theloved tracks that'd be tellar

stellar even

the playlists are not a super important thing. if they're there I'd appreciate it,but if no I can live with it

might be ok to start fresh or osmething..

yeah, I don't think anyone would delete your account without telling you...so that is concerning if you didn't do it...

it was that weird bug, i attempted to delete listens (as it) but that didn't work

I did attempt to delet the account (while it said i was logged in as it, but that a) didn't work and my own wasperectly fine after...)

zas: Stopping SIR again as Solr is timing out for some time.

as an asside. the "Connect to your SoundCloud account to play music on ListenBrainz." is giving an error on sSC: "Something went wrong (redirect_uri_mismatch)

The redirection URI provided does not match a pre-registered value."

oh uh. my followers/following got nuked too apaprently. uh

> I did attempt to delet the account

That was a bad idea, this means you have deleted your own account.

Deletions in LB are not instantaneous, they get queued and happen at the next full hour.

o. i see

lucifer: can you restore this recording_feedback data for ApeKattQuest, MonkeyPython when you have time? https://gist.github.com/mwiencek/4d85b963a6f785...

I updated it to reference their new user_id

well it said I was logged in to as another user XD

sorry for the sound I was unaware. m'bad

If there is a bug somewhere, the last thing you should do is a potentially destructive action, you will never know what happens.

I was testing!

:(

well i've learned it the hard way!

(lucky that it wasn't worse)

well..

this is my first time deleting an account accidentally on a service :O after 20 yeears on the internet andcomputers!

that's not bad right? :DD

Still better than having a test suite delete your entire production DB, cough 😉

:D

[@yvanzo](https://matrix.to/#/@yvanzo:chatbrainz.org): seems like prod Solr close to a major failure. do you think it's feasible to switch to the new cluster next week, bearing in mind I am on holiday for 2 weeks immediately after?

* prod Solr is close to

I second this analysis, the old cluster is rather unstable since few weeks.

is it expected that search requests are still bit laggy/timeout? some hours ago everything was blazing fast 🙂

solr cluster failing again

😢

atj: I'm lowering rate limit again, I'll try to restore a stable behavior...

yvanzo: did you stop sir-prod? or was it moved elsewhere?

zas: I stopped it 2h ago.

I guess we'll need to move to new cluster very soon

I still don't get why some nodes are reading disk at high rate (1,7Gb/s since 5 hours for solr 4), that's just abnormal.