#metabrainz

/

March 21st 2025

      • jasje[m]
        <GautamShorewala[> "jasje: aerozol @aerozol:matrix...." <- Looks good!!
      • jasje[m]: Put the hue sound tile on top
      • jasje[m]: In explore
      • jasje[m]: Also, maybe we can have a startup dialog alerting users of this feature
      • GautamShorewala[
        <jasje[m]> "Also, maybe we can have a..." <- I think a general dialog which can be shown after every update would be a better option
      • <jasje[m]> "Also, maybe we can have a..." <- Which shows all the latest changes
      • jasje[m]
        <GautamShorewala[> "Which shows all the latest..." <- Yup a general mechanism but i feel it has to be backend driven
      • BrainzGit
        [listenbrainz-server] 14mAmanullah7 opened pull request #3233 (03master…devtemp): Fixed Nonetype error! https://github.com/metabrainz/listenbrainz-serv...
      • bitmap[m]
        lucifer: yvanzo: reosarevok: I'm around now, let me know when you'd like to start
      • reosarevok[m]
        I'm around :)
      • yvanzo[m]
        Same!
      • BrainzGit
        [troi-recommendation-playground] 14mayhem opened pull request #168 (03main…improve_http_fetching): Improve http fetching https://github.com/metabrainz/troi-recommendati...
      • reosarevok[m]
        lucifer: let us know when you're around since seems everyone else is here :)
      • bitmap[m]
        I started another reindex on hip last night (using the latest sir 2to3 branch), but only for recordings. cpu_count: 8 / mem_limit: "8gb" were set on the indexer, mq, and search containers
      • it took under 7 hours to index recordings this time, whereas my previous test with the python2.7 sir took 12 hours. lucifer do you still have the timings from your mbdata run on trille?
      • yvanzo[m]
        reosarevok: Would you like to release a new version of mbdata for the next schema change now that it is under MeB’s umbrella?
      • bitmap[m]
        grafana shows CPU was maxed out, system RAM usage hovered around 6.5GB
      • reosarevok[m]
        yvanzo: It would make sense, yes :)
      • yvanzo[m]
        Of course, that requires changes to be ready in an MBS branch at first.
      • reosarevok[m]
        Yeah, once we have the changes merged in MBS and they're being tested we can look into making sure mbdata is ready too
      • I'll happily look into that at the time, remind me I promised if I inevitably forget :p
      • yvanzo[m]
        Let’s create a ticket to remind of it.
      • reosarevok[m]
        Sure. Where does that go, should we have a new mbdata project or?
      • yvanzo[m]
        A SEARCH ticket for updating SIR would do.
      • lucifer[m]
        I am around now
      • Don't have timings [@bitmap:chatbrainz.org](https://matrix.to/#/@bitmap:chatbrainz.org) but I think we'll be running it once again with memory constraints once I rearrange the commits.
      • reosarevok[m]
        yvanzo: SEARCH-740 - fill as needed if you can
      • BrainzBot
        SEARCH-740: Update mbdata / Sir to MBS schema 30 (2025 Q2) https://tickets.metabrainz.org/browse/SEARCH-740
      • lucifer[m]
        The performance improvements are expected because I fixed some lazy loaded queries in the branch. However last time we tested at least for recording core the cpu usage was excessive
      • <bitmap[m]> "I started another reindex on hip..." <- Oh noice
      • Let me fix the branch by tonight and we can run a full reindex on all cores over the weekend
      • yvanzo[m]
        bitmap: Can we add more constraints to the Compose file for testing resources?
      • You’ve set 8 CPUs, is it assuming 2 threads per CPU?
      • bitmap[m]
        that's what I was assuming, hip has 8 cores
      • yvanzo[m]
      • bitmap[m]
        during my test the CPU was completely maxed out with cpu_count: 8
      • yvanzo[m]
        Oh and swap has to be turned off for Solr container! https://docs.docker.com/reference/compose-file/...
      • bitmap[m]
        I had also set mem_limit: "8gb" on the sir, search, and mq containers during the test (didn't change swap settings)
      • yvanzo[m]
      • bitmap[m]
        is setting mem_swappiness on the search container not sufficient?
      • and memswap_limit
      • yvanzo[m]
        Yes it should be
      • bitmap[m]
        yvanzo: so I currently have https://gist.github.com/mwiencek/db5cdc647ece9d... any suggestion for tweaking the `cpu_count` / `mem_limit` values before I start another run?
      • rustynova[m]
        Made 2 new PRS on Melba to cleanup/update dependencies.
      • Before the second is merged, should we add dependabot as well? Or something similar?
      • bitmap[m]
        bitmap[m]: hip has 8 cores so I'm not sure cpu_count: 8 actually does anything
      • yvanzo[m]
        bitmap: Yes you can use sir-dev instead
      • (unless it is altering perfs?)
      • bitmap: Good point, cpu_count should be set to 2 then.
      • bitmap[m]
        ah, I wasn't quite sure of the difference with using sir-dev instead. it looks like the sir-dev Dockerfile is also using py2.7, so I'd have to make the same modifications to get the 2to3 branch working?
      • yvanzo[m]
      • bitmap[m]
        thanks, modifying .env does seem easier :)
      • ok, I started a new run on hip with sir-dev and and cpu_count: 2 on all three containers. swap disabled on solr container
      • yvanzo[m]
        Thanks!
      • lucifer[m]
        bitmap: hi! can you cancel it and pull the latest branch run it using that
      • bitmap[m]
        I'm on b5cd08037f4c8caa086a8d16e6b33d1d273d622c and git pull says already up-to-date, is that right?
      • lucifer[m]
        sorry i had forgotten to push, please check again
      • bitmap[m]
        nvm I see it now
      • running with the new code now
      • docker exec musicbrainz-docker-indexer-1 tail -f sir-reindex-2025-03-21.log if you want to check progress
      • BrainzGit
        [musicbrainz-server] 14mwiencek opened pull request #3503 (03master…rm-unused-code): Remove some unused code https://github.com/metabrainz/musicbrainz-serve...