#metabrainz

/

      • lucifer
        monkey: can you share console or request logs?
      • 2021-08-30 24205, 2021

      • monkey
        Yep, on it :)
      • 2021-08-30 24217, 2021

      • lucifer
        thanks!
      • 2021-08-30 24222, 2021

      • lucifer
        monkey: i just tried writing a review from test.lb and it works. can you try from it? maybe its some localhost issue
      • 2021-08-30 24254, 2021

      • akshaaatt[m]
      • 2021-08-30 24234, 2021

      • yvanzo
      • 2021-08-30 24241, 2021

      • akshaaatt[m]
        Neat. Thanks for sharing!
      • 2021-08-30 24222, 2021

      • alastairp
        btw, hi monkey!
      • 2021-08-30 24225, 2021

      • alastairp
        how are you?
      • 2021-08-30 24216, 2021

      • monkey
        Hi alastairp ! I am quite happy to be back where the sun doth shine, although today might not be the best example of that.
      • 2021-08-30 24231, 2021

      • monkey
        lucifer: Thanks for deloying, I'll test there
      • 2021-08-30 24243, 2021

      • alastairp
        it's nice to finally have some wet
      • 2021-08-30 24201, 2021

      • alastairp
        monkey: it'd be nice to get your feedback on https://github.com/metabrainz/listenbrainz-server… when you have time
      • 2021-08-30 24222, 2021

      • monkey
        Yep, it's on my list, thanks for the reminder
      • 2021-08-30 24229, 2021

      • alastairp
        fanks
      • 2021-08-30 24232, 2021

      • alastairp
        semi-related to that PR, I found LB-925
      • 2021-08-30 24233, 2021

      • BrainzBot
        LB-925: Support control/cmd click on react links to open links in a new tab https://tickets.metabrainz.org/browse/LB-925
      • 2021-08-30 24214, 2021

      • alastairp
        I can't remember if we had this same discussion back when you fixed LB-864
      • 2021-08-30 24215, 2021

      • BrainzBot
        LB-864: Newer/Older buttons do not work as links https://tickets.metabrainz.org/browse/LB-864
      • 2021-08-30 24230, 2021

      • monkey
        Ah, nice !
      • 2021-08-30 24203, 2021

      • monkey
        The ctrl+ part I didn't have a fix for, but it does look pretty easy to fix
      • 2021-08-30 24228, 2021

      • monkey
        There's also those right/left arrow key navigation events (to go to prev/next page) that shouldn't fire if an input is currently focused.
      • 2021-08-30 24232, 2021

      • alastairp
        yeah, if we can do a cross-browser/OS test to check that it works as expected in all combinations then I think it'd be an easy fix and will properly restore expected functionality :)
      • 2021-08-30 24206, 2021

      • monkey
        You end up navigation right and left when writing text in the pin listen or CB review modals
      • 2021-08-30 24224, 2021

      • alastairp
        I didn't even know about those events. I wonder if there's a way that we can expose this info better. perhaps a `?` help overlay like on other sites (github)
      • 2021-08-30 24241, 2021

      • alastairp
        oh, that doesn't sound ideal. I guess it just captures the event globally on the page?
      • 2021-08-30 24241, 2021

      • yvanzo
        reosarevok: Spotted a bug in beta thanks to Sentry: MBS-11923
      • 2021-08-30 24241, 2021

      • BrainzBot
        MBS-11923: Beta: Error: Props <script> for div.release-events-container not found. https://tickets.metabrainz.org/browse/MBS-11923
      • 2021-08-30 24207, 2021

      • reosarevok
        yvanzo: yeah, I saw, thanks - I opened two PRs for beta issues already, I can look into this one too
      • 2021-08-30 24223, 2021

      • monkey
        It does. Although looking at the code it seems I tried something that isn't working: https://github.com/metabrainz/listenbrainz-server…
      • 2021-08-30 24219, 2021

      • monkey
        Ah, of course "input" != "textarea" , that's an easy fix
      • 2021-08-30 24226, 2021

      • monkey
        OK, that makes a nice PR for me with a couple of tweaks, and I'll test those out on various platforms and browsers.
      • 2021-08-30 24228, 2021

      • yvanzo
        reosarevok: I was initially looking for hints about the “Oom kill detected” alert but found this instead.
      • 2021-08-30 24251, 2021

      • reosarevok
        Yeah, sorry, didn't look into those messages yet because I was looking at the beta issues chaban reported :)
      • 2021-08-30 24217, 2021

      • yvanzo
        bitmap, reosarevok: Digged both container logs and sentry for the 3rd OOM kill detected, and found nothing either.
      • 2021-08-30 24211, 2021

      • yvanzo
        Each time, all node processes but 1 were older than the alert.
      • 2021-08-30 24232, 2021

      • reosarevok
        The two warnings there on https://gist.github.com/yvanzo/130e392a5210de7455… are very old so I assume unrelated, not sure about the trace
      • 2021-08-30 24243, 2021

      • reosarevok
        (I really need to try and look into that email one again I think)
      • 2021-08-30 24209, 2021

      • yvanzo
        Most probably unrelated indeed.
      • 2021-08-30 24242, 2021

      • yvanzo
        (Because nothing can be found for the two other alerts.)
      • 2021-08-30 24255, 2021

      • monkey
        lucifer: Still getting CORS error trying to publish a CB review from test.LB
      • 2021-08-30 24206, 2021

      • monkey
        Same situation as before it looks like (OPTIONS request appears afterwards and I see no response headers in the failed POST request)
      • 2021-08-30 24210, 2021

      • reosarevok
        yvanzo: looking at the sentry error on that ticket, it's predated by
      • 2021-08-30 24212, 2021

      • reosarevok
      • 2021-08-30 24218, 2021

      • reosarevok
        That doesn't seem to be ours - userscript?
      • 2021-08-30 24232, 2021

      • reosarevok
        I expect chaban would have reported this to us otherwise
      • 2021-08-30 24200, 2021

      • reosarevok
        I'll tag them on the ticket and ask
      • 2021-08-30 24234, 2021

      • lucifer
        monkey: huh! weird. i have another hunch try disconnecting and reconnecting CB from test.lb
      • 2021-08-30 24245, 2021

      • monkey
        lucifer: Good hunch! Working for me now
      • 2021-08-30 24256, 2021

      • monkey
        Thanks for your help
      • 2021-08-30 24205, 2021

      • lucifer
        cool, i know the issue now.
      • 2021-08-30 24216, 2021

      • lucifer
        CORS headers are not applied to 40X requests.
      • 2021-08-30 24235, 2021

      • lucifer
        s/40X/50X
      • 2021-08-30 24244, 2021

      • lucifer
        somewhere in the request, an ISE occurs and a 500 is returned without cors headers so it looks like it is a cors issue but its actually an ISE.
      • 2021-08-30 24229, 2021

      • lucifer
        there's an issue with test.lb/beta.lb and lb setup. unlike spotify, CB does not support multiple redirect URIs. so each site has its own app.
      • 2021-08-30 24207, 2021

      • lucifer
        so if you register using beta and try to write a review from prod, the access token will not be considered valid. CB currently does not handle this error correctly, ending up in a 500.
      • 2021-08-30 24216, 2021

      • alastairp
        did the 500 happen on OPTIONS, or on the POST?
      • 2021-08-30 24235, 2021

      • lucifer
        POST methinks
      • 2021-08-30 24234, 2021

      • monkey
        Ahaa, good sleuthing
      • 2021-08-30 24251, 2021

      • bitmap
        yvanzo: weird that they're all on patton
      • 2021-08-30 24256, 2021

      • bitmap
        I'm confused though, from the graphs the memory increase happened in the webservice containers? https://stats.metabrainz.org/d/000000080/containe…
      • 2021-08-30 24211, 2021

      • bitmap
        there's no node executable inside those containers
      • 2021-08-30 24235, 2021

      • alastairp
        does anyone know who a transifex maintainer is?
      • 2021-08-30 24249, 2021

      • reosarevok
        alastairp: I'm one IIRC?
      • 2021-08-30 24250, 2021

      • alastairp
        I'm trying to update CB pot but it tells me "Only maintainers are allowed to update the source file."
      • 2021-08-30 24202, 2021

      • reosarevok
        Dunno if that's per-project tho
      • 2021-08-30 24209, 2021

      • alastairp
        reosarevok: can you try and make me one? alastairp
      • 2021-08-30 24243, 2021

      • reosarevok
        alastairp: I *think* I did?
      • 2021-08-30 24257, 2021

      • reosarevok
        Check
      • 2021-08-30 24204, 2021

      • alastairp
        I just did a push and it didn't complain at me!
      • 2021-08-30 24205, 2021

      • alastairp
        thanks
      • 2021-08-30 24245, 2021

      • bitmap
        akshaaatt[m]: thanks for the update! I'll try to find relevant tickets to assign you. I'm not sure we have ones for the homepage yet
      • 2021-08-30 24241, 2021

      • bitmap
        & I saw you posted mock-ups on the forums already today, cool
      • 2021-08-30 24258, 2021

      • reosarevok
        bitmap: so, how much do you know about sir / sqlalchemy?
      • 2021-08-30 24219, 2021

      • reosarevok
        I saw you fixed the last thing here, so :D
      • 2021-08-30 24231, 2021

      • bitmap
        I worked with it some years ago, so a bit, but a lot has probably escaped my memory
      • 2021-08-30 24234, 2021

      • bitmap
        but can try to help to the best of my ability :)
      • 2021-08-30 24245, 2021

      • reosarevok
        I was looking into the second issue marked on https://github.com/metabrainz/sir/pull/111
      • 2021-08-30 24213, 2021

      • reosarevok
        While doing that, I saw that there's now an actual intended way to defer all but some columns, load_only
      • 2021-08-30 24239, 2021

      • reosarevok
      • 2021-08-30 24247, 2021

      • yvanzo
        reosarevok: that could help you with: https://chatlogs.metabrainz.org/brainzbot/metabra…
      • 2021-08-30 24219, 2021

      • reosarevok
        Oh, I'll check that too, thanks yvanzo
      • 2021-08-30 24232, 2021

      • bitmap
        what's the second issue? "AttributeError: _wildcard_token"?
      • 2021-08-30 24237, 2021

      • reosarevok
        Yeah
      • 2021-08-30 24239, 2021

      • reosarevok
        But while trying to change that, it clearly is trying to load a relationship as column or something like that
      • 2021-08-30 24247, 2021

      • reosarevok
        So I get this (with the debugs added on that PR)
      • 2021-08-30 24252, 2021

      • reosarevok
        (eh, commit, no PR yet)
      • 2021-08-30 24207, 2021

      • reosarevok
      • 2021-08-30 24228, 2021

      • reosarevok
        The PoolWorker-1 2021-08-30 16:08:31,345 ERROR: can't locate strategy for <class 'sqlalchemy.orm.relationships.RelationshipProperty'> (('deferred', False), ('instrument', True)) bit is the actual issue, anyway
      • 2021-08-30 24253, 2021

      • lucifer
      • 2021-08-30 24237, 2021

      • MrClon
        There is some "add new area" requests in tracker. Can someone add them?
      • 2021-08-30 24242, 2021

      • reosarevok
        Oh, ffs
      • 2021-08-30 24251, 2021

      • reosarevok
        "instrument" isn't connected to MB instruments?
      • 2021-08-30 24254, 2021

      • reosarevok
        Well then :D
      • 2021-08-30 24256, 2021

      • lucifer
        yes lol
      • 2021-08-30 24201, 2021

      • reosarevok
        thanks for that hint at least, lucifer
      • 2021-08-30 24213, 2021

      • lucifer
        i was too confused when i saw that issue :D
      • 2021-08-30 24227, 2021

      • reosarevok
        MrClon: they'll get added eventually. Unless there's a huge urgency with any of them (which seems unlikely for AREQ) please just be patient :)
      • 2021-08-30 24251, 2021

      • reosarevok
        We do have an "instrument" relationship, so I kept trying to figure out why it was being hit
      • 2021-08-30 24230, 2021

      • bitmap
        being not-so-familiar with sqlalchemy I'm not sure what _get_strategy does or where the RelationshipProperty comes from
      • 2021-08-30 24219, 2021

      • reosarevok
        bitmap, yvanzo: re SEARCH-609 from that previous talk last year, is there any reason we never implemented any of the suggestions?
      • 2021-08-30 24220, 2021

      • BrainzBot
        SEARCH-609: SIR triggers do not work as intended when running a MusicBrainz Slave instance https://tickets.metabrainz.org/browse/SEARCH-609
      • 2021-08-30 24243, 2021

      • bitmap
        reosarevok: no, I don't think so. though incidentally I was working on something that could help here
      • 2021-08-30 24254, 2021

      • bitmap
        I'd like us to remove the need for these triggers + rabbitmq, and I think the best way to do that is have sir consume the dbmirror_pendingdata table
      • 2021-08-30 24214, 2021

      • bitmap
        but we'll need to change dbmirror so that it stores the old row versions
      • 2021-08-30 24221, 2021

      • aerozol has quit
      • 2021-08-30 24231, 2021

      • aerozol joined the channel
      • 2021-08-30 24204, 2021

      • lucifer
        reosarevok: can you log and share the list of required columns passed to the defer_everything_but function? i'd do it myself but my mb setup isn't up and running yet.
      • 2021-08-30 24222, 2021

      • bitmap
        pushed what I was working on to https://github.com/metabrainz/dbmirror2 (I started it a few years ago), but we'll need a plan to test and migrate to this
      • 2021-08-30 24235, 2021

      • reosarevok
        lucifer: I'm already logging those :)
      • 2021-08-30 24237, 2021

      • reosarevok
      • 2021-08-30 24241, 2021

      • reosarevok
        (removed other logging)
      • 2021-08-30 24248, 2021

      • reosarevok
        (well, most other)
      • 2021-08-30 24209, 2021

      • reosarevok
        It breaks after
      • 2021-08-30 24210, 2021

      • reosarevok
        PoolWorker-1 2021-08-30 16:48:33,635 DEBUG: Loading only ['gid', 'name', 'id'] on <class 'mbdata.models.InstrumentType'>
      • 2021-08-30 24220, 2021

      • reosarevok
        But I'm not sure if that's where the error is, or if it just does all of them, then breaks
      • 2021-08-30 24233, 2021

      • lucifer
        oh! i missed that :p. from the issue, it seems that it doesn't work on relationships. many of these are probably relationships but why did it work in the earlier way then...
      • 2021-08-30 24257, 2021

      • reosarevok
        In theory, we split relationships
      • 2021-08-30 24202, 2021

      • reosarevok
        But PoolWorker-1 2021-08-30 16:48:33,634 DEBUG: Loading only ['count', 'tag', 'instrument'] on <class 'mbdata.models.InstrumentTag'>
      • 2021-08-30 24208, 2021

      • reosarevok
        instrument there does seem to be a relationship
      • 2021-08-30 24244, 2021

      • reosarevok
        See if isinstance(prop, RelationshipProperty): on searchentities.py build_entity_query
      • 2021-08-30 24254, 2021

      • reosarevok
        My understanding is that this should detect and split relationships
      • 2021-08-30 24232, 2021

      • reosarevok
        Or is that only about composites, I guess
      • 2021-08-30 24234, 2021

      • reosarevok
        Anyway, it does seem to do the loads differently for relationships...
      • 2021-08-30 24235, 2021

      • lucifer
        not sure. but i would like to see understand why we need the eager loading anyways
      • 2021-08-30 24223, 2021

      • reosarevok
        I would like to understand why we need sqlalchemy :p
      • 2021-08-30 24235, 2021

      • reosarevok
        (ok, ok, apparently because otherwise we can't use mbdata so it's a full rewrite)
      • 2021-08-30 24222, 2021

      • reosarevok
        Oh, next meeting is Sep 9, anyway, lol. I should read the topic
      • 2021-08-30 24255, 2021

      • lucifer
        Sep 6 :)
      • 2021-08-30 24200, 2021

      • reosarevok
        oh. Yeah. That.
      • 2021-08-30 24236, 2021

      • reosarevok
        Given the weird way we use sqlalchemy, according to 2020 bitmap at least, wonder if we could change it to at least use it less weirdly
      • 2021-08-30 24254, 2021

      • lucifer
        i am generally in favor of using an ORM like sqlalchemy but if we have to resort weird stuff then its questionable
      • 2021-08-30 24225, 2021

      • lucifer
        indeed. i suggest maybe revert this commit for now and just fix the breaking changes?
      • 2021-08-30 24252, 2021

      • lucifer
        once we are on the latest version, we can reduce the weirdness. or at least document why it is necessary
      • 2021-08-30 24248, 2021

      • reosarevok
        I mean, I was seeing if this would avoid the breaking changes :p
      • 2021-08-30 24257, 2021

      • reosarevok
        But it just broke something else
      • 2021-08-30 24246, 2021

      • lucifer
        oh!
      • 2021-08-30 24258, 2021

      • reosarevok
        It's not like the error we had before,
      • 2021-08-30 24206, 2021

      • reosarevok
      • 2021-08-30 24210, 2021

      • reosarevok
        is any clearer to me :D
      • 2021-08-30 24237, 2021

      • lucifer
        ah, but it is the same issue i think.
      • 2021-08-30 24253, 2021

      • lucifer
        that issue i referred above mentioned about a wildcard being used internally
      • 2021-08-30 24257, 2021

      • reosarevok
        Might very well be, yeah
      • 2021-08-30 24207, 2021

      • bitmap
        it does seem wise to be on the latest sqlalchemy before we attempt any fix
      • 2021-08-30 24228, 2021

      • reosarevok
        bitmap: IIRC the idea was to fix stuff that broke in migration
      • 2021-08-30 24244, 2021

      • reosarevok
        So I assume this currently works in 1.0 but breaks in 1.1 for... reasons
      • 2021-08-30 24207, 2021

      • reosarevok
        Maybe you can find something in https://docs.sqlalchemy.org/en/14/changelog/migra… that helps you figure out why?
      • 2021-08-30 24212, 2021

      • bitmap
        right, but what if it breaks again in 1.2 or 1.3
      • 2021-08-30 24258, 2021

      • reosarevok
        This reads like Russian to me - I recognize some small bits, but the general text is still incomprehensible :D
      • 2021-08-30 24212, 2021

      • reosarevok
        bitmap: Hmm, well, dunno. I tried this because it was yvanzo's original approach