[listenbrainz-server] 14Singh-Nitin-15 opened pull request #3461 (03master…lb-1788): Removed calendar icon overlapping date labels in Fresh Releases timeline https://github.com/metabrainz/listenbrainz-server…
2025-12-21 35520, 2025
d4rkie_ joined the channel
2025-12-21 35510, 2025
d4rkie has quit
2025-12-21 35541, 2025
tanishkaarora joined the channel
2025-12-21 35504, 2025
tanishkaarora has quit
2025-12-21 35555, 2025
Kladky joined the channel
2025-12-21 35550, 2025
zas[m] has quit
2025-12-21 35549, 2025
pite_ joined the channel
2025-12-21 35555, 2025
pite has quit
2025-12-21 35523, 2025
petitminion joined the channel
2025-12-21 35522, 2025
petitminion has quit
2025-12-21 35558, 2025
petitminion joined the channel
2025-12-21 35508, 2025
petitminion has quit
2025-12-21 35521, 2025
mayhem[m]
zas: can you please help me setup the mapping testing server as normal (temp) server part of our infra? I don't want to keep having to fight the proxy and live behind our performance proxies.
2025-12-21 35520, 2025
mayhem[m]
sure. thanks!
2025-12-21 35522, 2025
zas[m] joined the channel
2025-12-21 35522, 2025
zas[m]
sure, but not now, can it wait tomorrow?
2025-12-21 35538, 2025
zer0bitz- has quit
2025-12-21 35548, 2025
zer0bitz- joined the channel
2025-12-21 35504, 2025
zas[m]
Do you feel better?
2025-12-21 35536, 2025
mayhem[m]
yes. I've been on a liquid diet for a week now and things are improving...
2025-12-21 35558, 2025
mayhem[m]
I've been able to work on the new mapper and that has been really well received.
2025-12-21 35514, 2025
mayhem[m]
now I want to see if I can get 300 reqs/sec from it.
Apart from the whole torrent backup thing, some really cool graphs
2025-12-21 35503, 2025
AkashDeep[m] joined the channel
2025-12-21 35503, 2025
AkashDeep[m]
Hi everyone,
2025-12-21 35503, 2025
AkashDeep[m]
I’d like to introduce Akash Deep, a 4th-year student who is interested in contributing and learning through GSoC. He’s eager to work on real-world open source problems and grow as a developer. Looking forward to learning and collaborating with the community.
2025-12-21 35507, 2025
AkashDeep[m]
I was checking GitHub for issues to work on, but I couldn’t find any open ones. Could you please let me know where I should look or how to get started?
NOTICE: LB-1881 (https://tickets.metabrainz.org/browse/LB-1881): Skipping a track while brainzplayer is searching a track forcefully skip everything
2025-12-21 35503, 2025
failure[m]
hey @reosarevok:chatbrainz.org just wanted to let you know; it would be awesome if you added me to the highly official Official List™ for this week's meeting and the weeks after as well :D
2025-12-21 35500, 2025
failure[m]
thanks :D
2025-12-21 35509, 2025
mayhem[m] uploaded an image: (420KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/chatbrainz.org/wbiRpRiycmAhmhWxGsnMSTIz/image.png >
right now. note the no https. data is from a few days ago.
2025-12-21 35532, 2025
julian45[m]
but i only was able to put in some of the preconditions for it last night, hence my question about db recency
2025-12-21 35538, 2025
julian45[m]
ah
2025-12-21 35504, 2025
mayhem[m]
you need something very recent?
2025-12-21 35526, 2025
mayhem[m]
I can try and replicate the server to current and build a new index in a bit.
2025-12-21 35549, 2025
dabeglavins682 has quit
2025-12-21 35554, 2025
julian45[m]
need is a strong word, but basically i forgot to put an alias on something (and thus my test case actually visible to the mapper) until last night
2025-12-21 35524, 2025
julian45[m]
no rush, if you’re already planning to rebuild for another reason later i can wait
2025-12-21 35533, 2025
mayhem[m]
let me finish a few things and see about getting fresher data.
2025-12-21 35520, 2025
petitminion has quit
2025-12-21 35509, 2025
jfjeiaj has quit
2025-12-21 35521, 2025
julian45[m]
sounds good
2025-12-21 35551, 2025
Jadesheher[m]
<mayhem[m]> "💥 holy shit, python is SLOW." <- fr, the python services I run take orders of magnitude more resources for much less load than even the Go services
2025-12-21 35545, 2025
mayhem[m]
Jadesheher[m]: I'm being reminded of that. holy hell.
2025-12-21 35550, 2025
mayhem[m]
but this shit needs to be fast, so fine.
2025-12-21 35517, 2025
julian45[m]
also, small ask for later when you’re not slogging through test cases and other more important things: instrumentation for logs/metrics/traces gathering would be excellent - opentelemetry or prometheus would be neat for this (though perhaps traces *might* be more data than needed given the sheer volume of requests that will likely be pushed through this thing)
2025-12-21 35518, 2025
julian45[m]
i know otel has cpp stuff available, prom idk
2025-12-21 35515, 2025
Jadesheher[m]
I feel like this service is unlikely to be the source of distributed bugs
2025-12-21 35519, 2025
julian45[m]
tracing would probably be more useful when implemented across the whole of a project tbh bc the mapper is just one service of multiple involved in the project work
Ideally you'd be able to go 0.1%, 1%, 10% etc of real requests get redirected
2025-12-21 35555, 2025
Jadesheher[m]
At the load balancer level
2025-12-21 35533, 2025
Jadesheher[m]
And then to avoid being woken up you have a trigger on your error rate metrics that rolls that back to the original service
2025-12-21 35530, 2025
Jadesheher[m]
mayhem[m]: That is also 'unable to connect' for me
2025-12-21 35546, 2025
Jadesheher[m]
Which is weird, because I loaded it before
2025-12-21 35508, 2025
julian45[m]
methods aside, some metrics for understanding how stupid fast the mapper is handling user input [or not, i.e., error rates] (and tracking other things like how big the cache is, how long your db queries are taking), as well as at least surfacing operations that are erroring to somewhere seems prudent in the “relatively soon after the worst of the dirty work is done” term imo
2025-12-21 35541, 2025
mayhem[m]
ok.
2025-12-21 35511, 2025
julian45[m]
* methods aside, some metrics for understanding how stupid fast the mapper is handling user input [or not, i.e., error rates] (and tracking other things like how big the cache is, how long your db queries are taking), as well as at least surfacing operations that are erroring to somewhere seems prudent in the “relatively soon after the dirty work is done” term imo
NOTICE: LB-1208 (https://tickets.metabrainz.org/browse/LB-1208): Remove "Last.fm Scrobbler" from Clients list
2025-12-21 35520, 2025
mayhem[m]
wow, its always easier to re-download and re-create the database than try and get the replication token set in musicbrainz-docker.
2025-12-21 35535, 2025
mayhem[m]
hmm. collecting performance stats flat out breaks my design model, julian45 . will need to think about what I can really do. but the whole point is to have a pile of threads screaming along with minimal coordination (just sharing per artist credit search indexes, nothing else).
2025-12-21 35559, 2025
Jadesheher[m]
Reverse proxy level performance is usually good enough
2025-12-21 35522, 2025
Jadesheher[m]
I know you can get some quite good metrics out of traefik
2025-12-21 35515, 2025
Jadesheher[m]
Failing that you can do coordination free structures because these stats are commutative
2025-12-21 35523, 2025
Jadesheher[m]
like hyperloglog if you want to get real fancy
2025-12-21 35502, 2025
mayhem[m]
that'll have to be it. I can have the main thread answer questions about number of threads and so....
2025-12-21 35546, 2025
mayhem[m]
oh. i just had an idea to maybe make this not horribly slow.
2025-12-21 35510, 2025
mayhem[m]
what is better to cappture, julian45 request rate or reporting requests handled?
2025-12-21 35523, 2025
mayhem[m]
I'll make a json admin view endpoint with stats.
2025-12-21 35548, 2025
Jadesheher[m]
Rate of requests is usually derived from total requests in the dashboard
2025-12-21 35506, 2025
Jadesheher[m]
mayhem[m]: Use the Prometheus format, it's what everything accepts
