#metabrainz

/

      • thejoker88140 joined the channel
      • Maxr1998_ joined the channel
      • thejoker8814 has quit
      • thejoker88140 is now known as thejoker8814
      • Maxr1998 has quit
      • zerodogg has quit
      • zerodogg joined the channel
      • yellowhatpro joined the channel
      • Kladky joined the channel
      • Kladky has quit
      • Kladky joined the channel
      • aerozol[m] joined the channel
      • aerozol[m]
        Loading this artist always gives me a server error on beta (works on prod): https://beta.musicbrainz.org/artist/869cf6cc-07...
      • relaxoMob joined the channel
      • reosarevok
        aerozol[m]: checking
      • zerodogg has quit
      • zerodogg joined the channel
      • aerozol[m] has quit
      • lucifer
        reosarevok: can look into it later
      • reosarevok
        Thanks!
      • BrainzGit
        [musicbrainz-server] 14reosarevok opened pull request #3249 (03master…react-18.3): Update React to 18.3.1 https://github.com/metabrainz/musicbrainz-serve...
      • reosarevok
        bitmap: ^ any suggestions on what else if anything needs to happen there?
      • Oh no, flow 235.1 brings 49 new errors :D
      • Due to "obj[key] where obj is not a dictionary object and key is a string used to not error and just silently return any. We now error on this pattern." it seems
      • Which we do a fair amount
      • BrainzGit
        [musicbrainz-server] 14reosarevok opened pull request #3250 (03master…flow-235): [WIP] Update to Flow 0.235.1 https://github.com/metabrainz/musicbrainz-serve...
      • atj
        yvanzo: ping me when you're around
      • Maxr1998_ has quit
      • reosarevok
        aerozol: hmm, works locally, but it's indeed broken in beta. I think there's some clash between beta and prod because of the changes in https://github.com/metabrainz/musicbrainz-serve... by derat
      • bitmap: ^
      • Also, the paste-a-date plugin has [Violation] Listener added for a 'DOMNodeInserted' mutation event. warnings in Chrome now, we should look into moving ahead with https://github.com/metabrainz/musicbrainz-serve...
      • bitmap: heh, I wonder if this means the weird comments are gone with react 18.3 for some reason: https://app.circleci.com/pipelines/github/metab...
      • Maxr1998 joined the channel
      • bitmap: oh, seems there's a ticket for the error aerozol found, MBS-13560
      • BrainzBot
        MBS-13560: Beta: Artist page not displayed anymore https://tickets.metabrainz.org/browse/MBS-13560
      • BrainzGit
        [musicbrainz-server] 14reosarevok merged pull request #3238 (03master…MBS-13533-fixes-yellowhatpro): MBS-13533: Fix order of cover art in release group artwork https://github.com/metabrainz/musicbrainz-serve...
      • d4rk joined the channel
      • d4rkie has quit
      • BrainzGit has quit
      • BrainzGit joined the channel
      • FichteFoll has quit
      • FichteFoll joined the channel
      • lucifer
        bitmap, yvanzo, reosarevok: reminder of oauth meeting tonight one hour prior to regular meeting time. mayhem also if you'd like to join.
      • mayhem will be around and will keep half an eye open for it.
      • BrainzGit
        [musicbrainz-server] 14yvanzo opened pull request #3251 (03production…mbs-13562): MBS-13562: Check editing unlimited strings https://github.com/metabrainz/musicbrainz-serve...
      • yvanzo
        bitmap, reosarevok ^ hopefully hotfixing production today
      • atj: around for a short time before going to lunch
      • Tarun_0x0 joined the channel
      • atj
        yvanzo: let me know when you're back
      • BrainzGit
        [troi-recommendation-playground] 14amCap1712 opened pull request #136 (03main…fix-imports): Fix troi.patch imports https://github.com/metabrainz/troi-recommendati...
      • Tarun_0x0 has quit
      • [troi-recommendation-playground] 14amCap1712 merged pull request #136 (03main…fix-imports): Fix troi.patch imports https://github.com/metabrainz/troi-recommendati...
      • [troi-recommendation-playground] release 03v-2024-04-29.0 has been published by 14amCap1712: https://github.com/metabrainz/troi-recommendati...
      • Tarun_0x0 joined the channel
      • outsidecontext
        mayhem: just read you sslcom blog post. Reminder me of checking the app signing for picard. So we must deal with this in October this year. I set myself a reminder.
      • derat joined the channel
      • mayhem
        ok, cool. anyone but ssl.com -- and we haven't gotten a lot of traction on this, but some people commented that this is common for ssl.com. sigh
      • reosarevok
        yvanzo: reviewing :)
      • relaxoMob has quit
      • derat
        reosarevok, aerozol: taking a look at that issue. it seems like load_related_info maybe isn't called in some cases: https://github.com/metabrainz/musicbrainz-serve...
      • relaxoMob joined the channel
      • reosarevok
        derat: the interesting thing is it loads just fine locally
      • That's why I thought maybe it was connected with a clash in the way things are stored in beta vs prod
      • For example, the new stuff from your load_related_info won't be in prod
      • If prod is supposed to cache the artist after the load_related_info call and then beta tries to access it for the same artist, it will be missing stuff
      • But I haven't read the code to check if that's the case
      • If it is the case, then I'm not sure there's a lot you can do other than put that bit of the load_related_info code in prod - we're probably doing a prod hotfix release today or tomorrow for the truncated name patch from yvanzo so we could also add this for you if needed
      • outsidecontext
        mayhem: yes, definitely not them. Unfortunately the last time I check other options were more expensive. But let's actually have a look at the situation in september / october, then we can evaluate the options.
      • mayhem
        good plan
      • reosarevok
        In that case we'd need to patch in the changes to load_related_info *and* the primary alias utils function
      • derat
        reosarevok: it sounds like there's a lot going on with prod vs. beta that i'm unaware of. :-) so it's the case that beta may use entities that have been loaded by prod?
      • reosarevok
        But let's maybe also check with bitmap once he's around :)
      • Yeah, they share a cache and we've had issues with this kind of thing before
      • That's why I'm guessing that's probably the issue
      • btw, aerozol on #mb said "Hmm, is anyone else on beta having trouble with the tracklist editor artist search? If I seed artists and then press the ‘search’ button, it won’t search, unless I delete a character or type something new into the field. Pressing ‘edit’ and then search works. If it’s not just me I will make a bug report"
      • Do you think this might be related to your PR too in some way?
      • derat
        reosarevok: i think this code needs more time to bake before it makes it to prod. what about a small patch to just avoid crashing in the case where the aliases are undefined?
      • reosarevok
        That would be a possibility too. Putting the code for calculating and loading aliases in prod seems relatively safe to me as well though, assuming this is the issue. But let's see what yvanzo and bitmap think
      • derat
        re tracklist artist search, hmm, it's not immediately obvious to me how this could be causing that. it'd be good to know if any JS errors are being logged. i'll see if i can repro that
      • reosarevok
        Yeah, it might be unrelated, just thought I'd ask :)
      • derat
        i wasn't able to repro the tracklist editor seach issue on my first attempt, so a ticket would probably be useful. when i seed a (single-artist) bandcamp release on beta, switch to the tracklist editor, and click the magnifying glass on a track credit, i get the usual spinner and autocomplete
      • discordbrainz
        <11sound.and.vision> hello musicbrainz friends.... this entity (Black) won't load in Beta https://beta.musicbrainz.org/artist/738457fb-62... throws error "Can't use an undefined value as an ARRAY reference..." etc loads fine in prod
      • derat
        re putting the load_related_info changes in prod, you have much more faith in my perl code than i do :-)
      • Tarun_0x0 has quit
      • reosarevok
        sound.and.vision: we were just talking about that issue :)
      • We'll try to have a fix today or tomorrow, but if need be please use prod for any artists you get that problem with for now
      • discordbrainz
        <11sound.and.vision> oh my bad, thank-you in advance 😄
      • BrainzGit
        [listenbrainz-server] 14amCap1712 opened pull request #2857 (03master…fix-tests): Fix tests https://github.com/metabrainz/listenbrainz-serv...
      • yellowhatpro has quit
      • bitmap
        reosarevok: derat: we just need to add the `aliases` Moose attribute to the artist class in prod, then clear the cache. my bad for not thinking of that during review
      • yvanzo
        reosarevok: Thanks!
      • derat
        bitmap: glad the fix sounds straightforward! is that something that you're able to do? :-)
      • yvanzo
        atj: Back!
      • bitmap
        derat: yea I will put a patch up soonish
      • reosarevok
        Thanks bitmap
      • thitmap
      • bitmap
        🤨
      • reosarevok
      • atj
        so bizarre that you've heard of look around you
      • derat
        yeah, thanks!
      • atj
        yvanzo: 2 things, zas has setup a separate LB for Sir, and I want to start indexing from the prod DB
      • reosarevok
        atj: I'm honestly not sure where this came from, maybe someone British from MB originally introduced me to it?
      • It was *AGES* ago in any case
      • atj
        yeah I figured it wasn't a recent thing
      • just odd how so many people randomly know very specific/ niche bits of British culture (particularly comedy)
      • probably over represented in geek culture tbf
      • reosarevok
        I mean, it has to be, because American geeks aren't funny
      • So you have to go search elsewhere :p
      • BrainzGit
        [musicbrainz-server] 14mwiencek opened pull request #3252 (03production…mbs-13560): MBS-13560: Sync `Entity::Artist` with beta https://github.com/metabrainz/musicbrainz-serve...
      • relaxoMob has quit
      • relaxoMob joined the channel
      • huhridge joined the channel
      • huhridge
        hi everyone! the meet's at 1900 cet right?
      • yvanzo
        huhridge: in 2h 40min from now :)
      • atj
        yvanzo: regarding switching to prod indexing, let me know once you're ready and i'll delete all shards / replicas from the cluster
      • yvanzo
        huhridge: Oops, in 1h 40min from now! 😬
      • huhridge
        yvanzo: thanks!
      • yvanzo
        atj: Before deleting, can we test the backup?
      • yellowhatpro joined the channel
      • atj
        yvanzo: how are backups performed? i see a HAProxy ACL for aretha that permits BACKUP commands
      • is there a script somewhere?
      • yvanzo
        atj: We can do that another time, with a sample database.
      • atj
        the main reason I want to get indexing from prod is to test the performance impact
      • yvanzo
        atj: They are performed by the `musicbrainz-search-indexes-dump` container on `aretha`.
      • atj
        i've completely disabled the document cache but commits will cause all the other caches to be flushed
      • yvanzo
        atj, bitmap, reosarevok: I unplugged SolrCloud from test.mb.o.
      • (It is back to using a standalone Solr instance.)
      • atj: will that be on the same nodes? Or on arm64 nodes already?
      • atj
        the installation been migrated to arm64 already
      • seeing approx. double the performance with 7 nodes
      • zas
        and it costs much less :)
      • yvanzo
        Ok, I’m setting up a separate SIR instance for performance tests. I will prepare the backup tests another time.
      • atj
      • *circuit breakers
      • i will try enabling the load average circuit breaker to be NUMCPUS + 2 or similar
      • i've noticed that sometimes certain VMs perform very poorly and the load average hits 80+
      • yvanzo
        By the way, a new version of Solr is available.
      • atj
        this actually slows down the entire cluster
      • yvanzo: thanks, I can test upgrading the cluster then :)
      • derat has quit
      • Tarun_0x0 joined the channel
      • reosarevok
        lucifer: hi! were we meeting now?
      • bitmap was about to ping everyone too
      • lucifer
        hi yes!
      • lol i forgot about it at the last minute myself. thanks for the ping.
      • let's wait for yvanzo and then start?