#metabrainz

/

      • yvanzo
        yyoung: for example: sudo docker-compose exec musicbrainz bash -c './script/create_test_db.sh && prove -l t/tests.t :: --tests EditArtistCredit'
      • 2021-05-24 14408, 2021

      • yyoung
        yvanzo: Yes I've tried those and they seems to be working, but now I need to fix the selenium tests
      • 2021-05-24 14422, 2021

      • yyoung
        I can try to use Jenkins CI, but that would be inefficient when I need to write new tests
      • 2021-05-24 14425, 2021

      • BrainzGit
        [musicbrainz-server] reosarevok opened pull request #2115 (master…consistent-license-block-2): Make Perl license block consistent (II) https://github.com/metabrainz/musicbrainz-server/…
      • 2021-05-24 14425, 2021

      • yvanzo
        Yes, it is the only immediately usable workaround.
      • 2021-05-24 14419, 2021

      • yyoung
        I see there's a musicbrainz-tests container being used by CI, will that work?
      • 2021-05-24 14437, 2021

      • yvanzo
        I tried adding a 'selenium' service based on https://github.com/SeleniumHQ/docker-selenium
      • 2021-05-24 14458, 2021

      • fnurl joined the channel
      • 2021-05-24 14429, 2021

      • yvanzo
        musicbraizn-tests is running all services inside the same container, whereas musicbrainz-docker has one container per service.
      • 2021-05-24 14452, 2021

      • yvanzo
      • 2021-05-24 14458, 2021

      • yyoung
        So is musicbrainz-tests open-source?
      • 2021-05-24 14408, 2021

      • yvanzo
        yyoung: yes, it is under docker/Dockerfile.tests in musicbrainz-server
      • 2021-05-24 14439, 2021

      • yvanzo
        It is not currently possible to run selenium.js within musicbrainz-docker because that script assumes everything runs in the same container.
      • 2021-05-24 14454, 2021

      • yyoung
        I see.
      • 2021-05-24 14433, 2021

      • BrainzGit
        [listenbrainz-server] amCap1712 opened pull request #1480 (master…do-not-sanitise): LB-767: stop 'sanitising' docker compose container names https://github.com/metabrainz/listenbrainz-server…
      • 2021-05-24 14455, 2021

      • bitmap
        yvanzo: I'm pretty sure it only requires sir in the same container (for access to the SQL scripts) -- the rest can be in other containers
      • 2021-05-24 14400, 2021

      • bitmap
        but luckily you don't need search stuff if you're only writing tests for the external links editor
      • 2021-05-24 14451, 2021

      • yvanzo
        bitmap: for example, proxy is set to localhost:5050
      • 2021-05-24 14420, 2021

      • bitmap
        well it has to run in the web container for sure
      • 2021-05-24 14457, 2021

      • bitmap
        but pg, redis, rabbitmq, etc. can all be in other containers
      • 2021-05-24 14404, 2021

      • yvanzo
        so it requires selenium in the same container?
      • 2021-05-24 14414, 2021

      • bitmap
        it requires chrome or firefox & a webdriver for either of those browsers
      • 2021-05-24 14430, 2021

      • bitmap
        in the web container
      • 2021-05-24 14438, 2021

      • bitmap
        another option for yyoung is to download an existing musicbrainz-tests image and mount his local musicbrainz-server checkout in it
      • 2021-05-24 14417, 2021

      • bitmap
        I do this when I have to debug jenkins failures locally
      • 2021-05-24 14418, 2021

      • yvanzo
        sudo docker pull metabrainz/musicbrainz-tests:2021-05 # I guess
      • 2021-05-24 14430, 2021

      • bitmap
        then docker run -d --name mbtest metabrainz/musicbrainz-tests:v-2021-05
      • 2021-05-24 14448, 2021

      • bitmap
        + something for mounting the volume, I forgot the option
      • 2021-05-24 14453, 2021

      • BrainzGit
        [musicbrainz-server] reosarevok merged pull request #2112 (master…sql-caps): Standardize (uppercase) caps on SQL strings https://github.com/metabrainz/musicbrainz-server/…
      • 2021-05-24 14409, 2021

      • BrainzGit
        [musicbrainz-server] mwiencek merged pull request #2111 (master…build-materialize-tables-separate-txns): BuildMaterializedTables: use separate transactions https://github.com/metabrainz/musicbrainz-server/…
      • 2021-05-24 14421, 2021

      • BrainzGit
        [musicbrainz-server] mwiencek merged pull request #2110 (master…mbs-11671): MBS-11671: force hourCycle: 'h23' for %H in formatUserDate https://github.com/metabrainz/musicbrainz-server/…
      • 2021-05-24 14431, 2021

      • BrainzGit
        [musicbrainz-server] mwiencek merged pull request #2107 (production…json-dumps-gpg-fix): Fix double JSON dump gpg-signing https://github.com/metabrainz/musicbrainz-server/…
      • 2021-05-24 14421, 2021

      • _lucifer
        the spark_result queue is moving slowly. unfortunately, logs are empty so no way to tell where it is taking so long.
      • 2021-05-24 14458, 2021

      • ruaok
        _lucifer: spark result queue is just saving the data to the DB no?
      • 2021-05-24 14457, 2021

      • _lucifer
        ruaok: yes, right. that's my understanding. however, it would be nice to know where it is currently at to know where we can speed up stuff.
      • 2021-05-24 14415, 2021

      • _lucifer
        we can probably add a few debug log statements here like we have elsewhere.
      • 2021-05-24 14417, 2021

      • _lucifer
        ruaok: alastairp: also i figured when this error happens, https://sentry.metabrainz.org/metabrainz/listenbr…. it happens when consul restarts the service don't know why though.
      • 2021-05-24 14445, 2021

      • _lucifer
        i matched the time of the errors with the time since the container was up and both were same.
      • 2021-05-24 14438, 2021

      • ruaok
        yeah, makes sense.
      • 2021-05-24 14421, 2021

      • ruaok
        _lucifer: the metric writer is finally stable and not losing data: https://stats.metabrainz.org/d/OGg5QUCGz/listenbr…
      • 2021-05-24 14441, 2021

      • _lucifer
        nice :D
      • 2021-05-24 14400, 2021

      • ruaok
        we could add some metrics for the write rate of the spark results.
      • 2021-05-24 14423, 2021

      • _lucifer
        !m ruaok
      • 2021-05-24 14423, 2021

      • BrainzBot
        You're doing good work, ruaok!
      • 2021-05-24 14431, 2021

      • ruaok
        ha. lol.
      • 2021-05-24 14435, 2021

      • _lucifer
        yup, that makes sense.
      • 2021-05-24 14443, 2021

      • ruaok hasnt been proud of his work in the last couple of weeks
      • 2021-05-24 14426, 2021

      • sumedh joined the channel
      • 2021-05-24 14426, 2021

      • _lucifer
        should we do the same for stats as we do for similar users?
      • 2021-05-24 14438, 2021

      • _lucifer
        i see why there are so slow, one message per user per stt
      • 2021-05-24 14440, 2021

      • _lucifer
        *stat
      • 2021-05-24 14459, 2021

      • ruaok
        >should we do the same for stats as we do for similar users?
      • 2021-05-24 14410, 2021

      • _lucifer
        currently we insert 1 stat, then process next.
      • 2021-05-24 14410, 2021

      • ruaok
        what do we do for similar users?
      • 2021-05-24 14418, 2021

      • ruaok
        oh geez.
      • 2021-05-24 14431, 2021

      • _lucifer
        one message, atomically replace tables on the fly.
      • 2021-05-24 14434, 2021

      • ruaok
        open a transaction and write 500 stats or so.
      • 2021-05-24 14443, 2021

      • _lucifer
        if not that at least batch
      • 2021-05-24 14445, 2021

      • _lucifer
        yes
      • 2021-05-24 14454, 2021

      • ruaok
        are we replacing the WHOLE table each time?
      • 2021-05-24 14410, 2021

      • ruaok
        if so, then yes. but if not, then add batch writing.
      • 2021-05-24 14418, 2021

      • _lucifer
        that was my understanding but you'd know better :)
      • 2021-05-24 14420, 2021

      • ruaok
        well, we should do batch writing in any case.
      • 2021-05-24 14440, 2021

      • ruaok
        let me think for a sec.
      • 2021-05-24 14447, 2021

      • ruaok
        how do stats work for users who stop submiting listens and disappear.
      • 2021-05-24 14405, 2021

      • ruaok
        do we still calculate their stats and then blast them into the DB?
      • 2021-05-24 14422, 2021

      • ruaok
        it really comes down to this: do we replace the whole table or not?
      • 2021-05-24 14441, 2021

      • ruaok
        if not, can we? if we can, then lets do the table swap trick -- it works rather quite well.
      • 2021-05-24 14447, 2021

      • _lucifer
        with batch writing, what would happen if one stat in the batch fails to insert?
      • 2021-05-24 14425, 2021

      • _lucifer
        i am not sure but i think we keep on calculating stats for users who disappear because that's what happens for other things in spark cluster
      • 2021-05-24 14448, 2021

      • ruaok
        > with batch writing, what would happen if one stat in the batch fails to insert?
      • 2021-05-24 14450, 2021

      • Sophist_UK has quit
      • 2021-05-24 14456, 2021

      • ruaok
        take a look at how the timescale writer writes listens.
      • 2021-05-24 14408, 2021

      • _lucifer
      • 2021-05-24 14409, 2021

      • _lucifer
        yeah
      • 2021-05-24 14421, 2021

      • _lucifer
        we continue calculating
      • 2021-05-24 14423, 2021

      • ruaok
        it does an insert and if there is a conflict, it discards those listens. it returns the IDs of all the rows that were inserted.
      • 2021-05-24 14451, 2021

      • _lucifer
        ah! makes sense.
      • 2021-05-24 14458, 2021

      • ruaok
        but for the table swap idea, you wouldn't even do that.
      • 2021-05-24 14415, 2021

      • ruaok
        having indexes on tables for bulk insert DRASTICALLY slows down the insert.
      • 2021-05-24 14423, 2021

      • ruaok
        you don't even run that in a transaction.
      • 2021-05-24 14436, 2021

      • ruaok
        let me put the steps in a gist for you.
      • 2021-05-24 14442, 2021

      • _lucifer
        great, thanks!
      • 2021-05-24 14450, 2021

      • _lucifer
        also, i wonder how much difference is there in pulling 10k small messages from rabbitmq than 1 relatively big message.
      • 2021-05-24 14405, 2021

      • ruaok
        I doubt that makes a diff.
      • 2021-05-24 14409, 2021

      • _lucifer
        cool, i'll need to make changes to spark side anyways to send just 1 message to keep stuff simpler on lemmy side.
      • 2021-05-24 14433, 2021

      • ruaok
      • 2021-05-24 14443, 2021

      • ruaok
        > cool, i'll need to make changes to spark side anyways to send just 1 message to keep stuff simpler on lemmy side.
      • 2021-05-24 14411, 2021

      • ruaok
        that is a HUGE amount of data to send in one message, I don't think that is a good idea. I would leave it as is for now. but for inserts, write 500 stats in one go.
      • 2021-05-24 14416, 2021

      • ruaok
        something configurable, at least.
      • 2021-05-24 14456, 2021

      • _lucifer
        yeah, right, the size would be too large. so let's modify spark so that each message has 500 stats or whatever the config is.
      • 2021-05-24 14412, 2021

      • ruaok
        on the gist, there is a tricky detail... indexes are not renamed after a table rename. so you need to rename them as well, which is an extra step. OR you can just give these indexes a unique name e.g. col1_col2_ndx_table_<timestamp> , where timestamp is just an epoch int time.
      • 2021-05-24 14423, 2021

      • ruaok
        this way you'll never have to rename indexes.
      • 2021-05-24 14438, 2021

      • ruaok
        > yeah, right, the size would be too large. so let's modify spark so that each message has 500 stats or whatever the config is.
      • 2021-05-24 14440, 2021

      • ruaok
        that works.
      • 2021-05-24 14459, 2021

      • _lucifer
        another issue is how do we know that all messages for stats have ended so that we know it is time to rename tables?
      • 2021-05-24 14417, 2021

      • ruaok
        pass a message with zero entries?
      • 2021-05-24 14448, 2021

      • _lucifer
        right, like a sentinel. that would probably work.
      • 2021-05-24 14403, 2021

      • ruaok
        yep
      • 2021-05-24 14433, 2021

      • _lucifer
        all stats are however stored in the same table, so we need the sentinel after the last stats type only.
      • 2021-05-24 14420, 2021

      • ruaok
        well, we will need to replace the WHOLE table for all stats types in one go.
      • 2021-05-24 14448, 2021

      • ruaok
        perhaps we need a "stats are a comin'" message and "stats be done" message.
      • 2021-05-24 14452, 2021

      • _lucifer
        right, so we need to wait for all stats to arrive.
      • 2021-05-24 14407, 2021

      • _lucifer
        yes that sounds better.
      • 2021-05-24 14433, 2021

      • _lucifer
        i think the rest of stuff should be similar to similar_users.
      • 2021-05-24 14437, 2021

      • ruaok
        do step 1 for the stats are coming. step 2 for the stats messages and step 3-6 for the stats are done message.
      • 2021-05-24 14452, 2021

      • ruaok
        that feels pretty right.
      • 2021-05-24 14416, 2021

      • _lucifer
        yes, much better than what we have currently.
      • 2021-05-24 14454, 2021

      • _lucifer
        thanks! i'll put all this in a ticket and assign to myself.
      • 2021-05-24 14415, 2021

      • ruaok
        sweet
      • 2021-05-24 14445, 2021

      • _lucifer
        queue is clear. reports should be updated.
      • 2021-05-24 14426, 2021

      • _lucifer
        some error is preventing the stats from loading for me. https://sentry.metabrainz.org/metabrainz/listenbr…
      • 2021-05-24 14453, 2021

      • _lucifer
        api shows stats have been updated.
      • 2021-05-24 14405, 2021

      • ruaok
        repors fail for me as well.
      • 2021-05-24 14416, 2021

      • ruaok
        TypeError: a[0] is undefined
      • 2021-05-24 14423, 2021

      • _lucifer
        Mr_Monkey: ^
      • 2021-05-24 14437, 2021

      • ruaok
        charts work, though.
      • 2021-05-24 14446, 2021

      • _lucifer
        yeah, same here.
      • 2021-05-24 14447, 2021

      • yvanzo has quit
      • 2021-05-24 14443, 2021

      • BrainzGit
        [musicbrainz-server] reosarevok opened pull request #2116 (master…MBS-11679): MBS-11679: Wish editors a happy MB anniversary https://github.com/metabrainz/musicbrainz-server/…
      • 2021-05-24 14426, 2021

      • yvanzo joined the channel
      • 2021-05-24 14453, 2021

      • akshaaatt[m]
        Hi! I have been planning to maintain a google doc https://bit.ly/3wIkukP where I will update my progress and add a few reminders/comments/anything of importance to it so that I am through with my work. The link is editable to all and you can add any suggestions for me as well. You could star this doc in case you want daily updates from my end 😇 Other than that what all do you expect from me by the end of the first
      • 2021-05-24 14453, 2021

      • akshaaatt[m]
        evaluation _lucifer ? According to the proposal that would fill me with the Tagger implementation, codebase documentation for it and the UI updates.
      • 2021-05-24 14446, 2021

      • _lucifer
        akshaaatt[m]: the doc looks super nice. however, I would suggest you to make a post on community.metabrainz.org and keep posting updates there (at least for the user facing improvements). this way the rest of community can remain up to date on the status of the work and make suggestion as well.
      • 2021-05-24 14436, 2021

      • akshaaatt[m]
        Okaay! That sounds great. So I'll formulate and make a new topic on the forum soon.
      • 2021-05-24 14427, 2021

      • _lucifer
        sounds good. regarding updates, i think once a week is fine, if possible preferably in the weekly meeting we have on monday. in case you cannot make it, you can always send your review to Freso before the meeting.
      • 2021-05-24 14426, 2021

      • _lucifer
        If you prefer with a more frequent timeline, feel free to tell its fine with me. and you can reach out any time you are stuck somewhere or would like to discuss something etc.
      • 2021-05-24 14423, 2021

      • akshaaatt[m]
        Thank you _lucifer ! So today is Monday right?! I'll attend my first meeting today then. I guess it starts in an hour from now?
      • 2021-05-24 14448, 2021

      • _lucifer
        yes right. awesome! :D
      • 2021-05-24 14417, 2021

      • TOPIC: MetaBrainz Community and Development channel | MusicBrainz non-development: #musicbrainz | Channel is logged; see https://musicbrainz.org/doc/IRC for details | Agenda (next meeting: 2021-05-24): Reviews, change of IRC network [ORG-49] (Freso)
      • 2021-05-24 14447, 2021

      • _lucifer
        ruaok: zeppelin setup on leader on port 9999. the UI looks a bit different but is similar to Jupyter Notebook. i connect to it as `ssh -L 9999:leader:9999 leader.listenbrainz.org`
      • 2021-05-24 14450, 2021

      • ZaphodBeeblebrox
        ORG-49
      • 2021-05-24 14451, 2021

      • BrainzBot
        ORG-49: Move IRC channels/presence off freenode https://tickets.metabrainz.org/browse/ORG-49
      • 2021-05-24 14406, 2021

      • ZaphodBeeblebrox is now known as CatQuest
      • 2021-05-24 14453, 2021

      • fnurl has quit
      • 2021-05-24 14412, 2021

      • Sophist-UK joined the channel
      • 2021-05-24 14455, 2021

      • Sophist-UK has quit
      • 2021-05-24 14422, 2021

      • Sophist-UK joined the channel
      • 2021-05-24 14449, 2021

      • reosarevok
        ruaok, alastairp or other EU-based electronic components and whatnot people: I would like to try something like https://calnature.org/blog/2017/9/27/diy-moth-lig…
      • 2021-05-24 14420, 2021

      • reosarevok
        Do you have any suggestions of shops with cheap postage inside Europe that would sell that sort of "DJ LEDs"?
      • 2021-05-24 14452, 2021

      • reosarevok
        AliExpress is full of that stuff but IIRC Estonia has started charging fairly big customs fees for Chinese imports
      • 2021-05-24 14421, 2021

      • CatQuest
        :o moths
      • 2021-05-24 14446, 2021

      • CatQuest
        HM. I would ask "clas ohlson" here hey'd have all the things.
      • 2021-05-24 14421, 2021

      • CatQuest
        does estonia not have some electrics/wire/gadget/electronics chain liek so?
      • 2021-05-24 14435, 2021

      • reosarevok
        They do, but they seem to mostly have visible spectrum stuff, although I might still go ask
      • 2021-05-24 14453, 2021

      • reosarevok
        yvanzo: I updated https://github.com/metabrainz/musicbrainz-server/…, good catch with my mis-squash
      • 2021-05-24 14430, 2021

      • legoktm waves
      • 2021-05-24 14438, 2021

      • CatQuest
        huh seems 365 mostly attracted not-moths but beetles and lacewing
      • 2021-05-24 14423, 2021

      • Freso
        <BANG>
      • 2021-05-24 14425, 2021

      • Freso
        It’s Asparagus Monday!