#metabrainz

/

      • samphippen
        you'd call reportscrobbles
      • 2015-09-18 26148, 2015

      • samphippen
        with the task again right?
      • 2015-09-18 26151, 2015

      • Craig has quit
      • 2015-09-18 26153, 2015

      • alastairp
        yeah
      • 2015-09-18 26158, 2015

      • samphippen
        as opposed to dispatch
      • 2015-09-18 26159, 2015

      • samphippen
        I mean
      • 2015-09-18 26100, 2015

      • samphippen
        at the end of the day
      • 2015-09-18 26106, 2015

      • samphippen
        they get deduplicated at the server side
      • 2015-09-18 26111, 2015

      • samphippen
        so it's probably not an issue right?
      • 2015-09-18 26113, 2015

      • Craig joined the channel
      • 2015-09-18 26114, 2015

      • alastairp
        well
      • 2015-09-18 26124, 2015

      • alastairp
        we have 2 clients sending us data as fast as they can
      • 2015-09-18 26129, 2015

      • alastairp
        and we’re responding with http400
      • 2015-09-18 26135, 2015

      • alastairp
        and it seems that they’re trying again
      • 2015-09-18 26145, 2015

      • weeksio_ has quit
      • 2015-09-18 26118, 2015

      • samphippen
        ok
      • 2015-09-18 26127, 2015

      • samphippen
        I'd put a settimeout
      • 2015-09-18 26128, 2015

      • alastairp
        I’m just concerned that the for loop in dispatch can be called from many callbacks, causing multiple submissions of data in toReport
      • 2015-09-18 26129, 2015

      • samphippen
        and retry in there
      • 2015-09-18 26141, 2015

      • samphippen
        really
      • 2015-09-18 26143, 2015

      • samphippen
        instead of looping it
      • 2015-09-18 26146, 2015

      • samphippen
        you should pop each item
      • 2015-09-18 26148, 2015

      • samphippen
        until it's empty
      • 2015-09-18 26148, 2015

      • samphippen
        then
      • 2015-09-18 26149, 2015

      • alastairp
        yeah
      • 2015-09-18 26154, 2015

      • alastairp
        sounds good
      • 2015-09-18 26155, 2015

      • ruaok
        one easy step is to not ever retry a submission that gets a 400 error.
      • 2015-09-18 26156, 2015

      • samphippen
        that case can't happen
      • 2015-09-18 26114, 2015

      • alastairp
        ok. what error handler will we see 400 in? onerror or onabort?
      • 2015-09-18 26121, 2015

      • samphippen
        literally no idea
      • 2015-09-18 26125, 2015

      • alastairp
        ok, thanks
      • 2015-09-18 26127, 2015

      • samphippen
        :P
      • 2015-09-18 26146, 2015

      • alastairp
        nice video, btw
      • 2015-09-18 26106, 2015

      • samphippen
        thanks
      • 2015-09-18 26107, 2015

      • samphippen
        :D
      • 2015-09-18 26124, 2015

      • samphippen
        it's a hard life, being in chicago, appearing on the BBC, speaking at conferences etc
      • 2015-09-18 26151, 2015

      • Bookzombie has quit
      • 2015-09-18 26104, 2015

      • ruaok
        say hi to your abused circadian rhythm for me. :)
      • 2015-09-18 26108, 2015

      • Bookzombie joined the channel
      • 2015-09-18 26159, 2015

      • Craig has quit
      • 2015-09-18 26107, 2015

      • alastairp
        I have no idea what I’m doing
      • 2015-09-18 26114, 2015

      • Craig joined the channel
      • 2015-09-18 26129, 2015

      • alastairp
        I would have thought that an onerror event would have a status field, or something
      • 2015-09-18 26129, 2015

      • Craig1 joined the channel
      • 2015-09-18 26157, 2015

      • Craig has quit
      • 2015-09-18 26124, 2015

      • github joined the channel
      • 2015-09-18 26124, 2015

      • github
        [listenbrainz-server] mayhem pushed 2 new commits to master: https://github.com/metabrainz/listenbrainz-server…
      • 2015-09-18 26124, 2015

      • github
        listenbrainz-server/master 6d8aa0c Robert Kaye: Do not throw 400 on empty payload
      • 2015-09-18 26124, 2015

      • github
        listenbrainz-server/master 1629807 Robert Kaye: Merge branch 'master' of github.com:metabrainz/listenbrainz-server
      • 2015-09-18 26124, 2015

      • github has left the channel
      • 2015-09-18 26148, 2015

      • chirlu`
        alastairp: jQuery?
      • 2015-09-18 26104, 2015

      • alastairp
        no, xmlhttprequest
      • 2015-09-18 26124, 2015

      • Craig1 has quit
      • 2015-09-18 26151, 2015

      • chirlu`
        It seems it has?
      • 2015-09-18 26110, 2015

      • chirlu`
        “XMLHttpRequest.status Read only
      • 2015-09-18 26110, 2015

      • chirlu`
        Returns an unsigned short with the status of the response of the request. This is the HTTP result code (for example, status is 200 for a successful request).”
      • 2015-09-18 26103, 2015

      • alastairp
        yeah, I found that finally
      • 2015-09-18 26110, 2015

      • alastairp
        so is that the argument of the callback?
      • 2015-09-18 26128, 2015

      • samphippen has quit
      • 2015-09-18 26116, 2015

      • chirlu`
        Not sure what the callback receives, but you can access the original object in any case.
      • 2015-09-18 26158, 2015

      • alastairp
        ah, ok
      • 2015-09-18 26121, 2015

      • chirlu`
        “
      • 2015-09-18 26121, 2015

      • chirlu`
        Invoke callback with five arguments, the first one having the value of E's message attribute, the second having the value of E's filename attribute, the third having the value of E's lineno attribute, the fourth having the value of E's colno attribute, the fifth having the value of E's error attribute, and with the callback this value set to E's currentTarget. Let return value be the callback's return value. [WEBIDL]”
      • 2015-09-18 26156, 2015

      • chirlu`
      • 2015-09-18 26104, 2015

      • UmkaDK_ has quit
      • 2015-09-18 26146, 2015

      • JesseW has quit
      • 2015-09-18 26127, 2015

      • fractalizator has quit
      • 2015-09-18 26132, 2015

      • github joined the channel
      • 2015-09-18 26132, 2015

      • github
        [listenbrainz-server] mayhem pushed 1 new commit to master: https://github.com/metabrainz/listenbrainz-server…
      • 2015-09-18 26132, 2015

      • github
        listenbrainz-server/master a031040 Robert Kaye: Use ujson for the kafka consumer
      • 2015-09-18 26132, 2015

      • github has left the channel
      • 2015-09-18 26158, 2015

      • alastairp
        russss: how do I update chef? (and therefore pull in pgbouncer)
      • 2015-09-18 26147, 2015

      • CatQuest
        uh hi, should i stop importing my stuff? it's me
      • 2015-09-18 26125, 2015

      • CatQuest
        in anycase, my browser is eating like 11 gigs already
      • 2015-09-18 26128, 2015

      • ruaok
        heh.
      • 2015-09-18 26140, 2015

      • ruaok
        we know of a bug, but have not pushed a fix yet
      • 2015-09-18 26102, 2015

      • alastairp
        well done to the developers of your browser
      • 2015-09-18 26116, 2015

      • russss
        alastairp: um, I'm not sure, RJ2 set that bit up and I don't know how he did it
      • 2015-09-18 26136, 2015

      • alastairp
        ok. we might have just apt-get install’d pgbouncer and copied the configuration in
      • 2015-09-18 26115, 2015

      • russss
        that's fine for the moment
      • 2015-09-18 26117, 2015

      • ruaok
        russss: we're still having some performance issues.
      • 2015-09-18 26122, 2015

      • ruaok
        got a moment to log in and look around?
      • 2015-09-18 26132, 2015

      • ruaok
        some 504s, but overall the site is very slow.
      • 2015-09-18 26140, 2015

      • ruaok
        and we just deployed pgbouncer.
      • 2015-09-18 26144, 2015

      • ruaok
        no improvement
      • 2015-09-18 26127, 2015

      • russss
        sure, you'll need to create me an account though :)
      • 2015-09-18 26136, 2015

      • ruaok
        ok.
      • 2015-09-18 26147, 2015

      • Craig joined the channel
      • 2015-09-18 26122, 2015

      • ruaok
        pmesg me an ssh key?
      • 2015-09-18 26128, 2015

      • russss
      • 2015-09-18 26126, 2015

      • Craig has quit
      • 2015-09-18 26132, 2015

      • ruaok
        russss: give it a shot
      • 2015-09-18 26140, 2015

      • ruaok
        user russ
      • 2015-09-18 26147, 2015

      • ruaok
      • 2015-09-18 26159, 2015

      • ruaok
        yay
      • 2015-09-18 26101, 2015

      • russss
        cool, I'm in
      • 2015-09-18 26125, 2015

      • russss
        holy uwsgi batman
      • 2015-09-18 26134, 2015

      • ruaok
        lol
      • 2015-09-18 26140, 2015

      • ruaok
        you now have sudo privs
      • 2015-09-18 26145, 2015

      • russss
        cool
      • 2015-09-18 26147, 2015

      • ruaok
        ok, I need to run off for food.
      • 2015-09-18 26152, 2015

      • ruaok
        thanks for helping russ. :)
      • 2015-09-18 26100, 2015

      • alastairp
        and I’m heading home, I’ll be back in 10
      • 2015-09-18 26112, 2015

      • ruaok has quit
      • 2015-09-18 26129, 2015

      • willejs joined the channel
      • 2015-09-18 26127, 2015

      • CatQuest
        alastairp: well it's literally cutting edge firefox so..
      • 2015-09-18 26138, 2015

      • CatQuest
        cutting as in my space is minche mear :P
      • 2015-09-18 26146, 2015

      • CatQuest
        meat*
      • 2015-09-18 26140, 2015

      • CatQuest
        had 500 mg left, turned it off and now i'e got 3 gigs back
      • 2015-09-18 26155, 2015

      • CatQuest
        3.8
      • 2015-09-18 26121, 2015

      • CatQuest
        so yea. complain to firefox (not that that will fix anything :()
      • 2015-09-18 26135, 2015

      • CatQuest
        now 4,88
      • 2015-09-18 26136, 2015

      • CatQuest
        yea
      • 2015-09-18 26150, 2015

      • CatQuest
        steadily releasing gigs at a time
      • 2015-09-18 26130, 2015

      • willejs has quit
      • 2015-09-18 26108, 2015

      • russss
        can't get anything obvious, I'm tempted to hook it up to newrelic temporarily as that'll sort it out
      • 2015-09-18 26127, 2015

      • russss
        although if I sign up for an account I'll have their sales people all over me for ever
      • 2015-09-18 26142, 2015

      • alastairp
        I’m not seeing any HTTP504 in the nginx logs
      • 2015-09-18 26102, 2015

      • alastairp
        so maybe it’s just settling down, or people aren’t submitting
      • 2015-09-18 26140, 2015

      • russss
        stuff's still slow
      • 2015-09-18 26157, 2015

      • russss
        check out "uwsgitop localhost:5050"
      • 2015-09-18 26127, 2015

      • russss
        (am just about to restart it)
      • 2015-09-18 26106, 2015

      • alastairp
        uwsgitop! I didn’t know that existed
      • 2015-09-18 26135, 2015

      • CatQuest
        I stopped submitting if that helps
      • 2015-09-18 26136, 2015

      • russss
        you have to install it from pip
      • 2015-09-18 26154, 2015

      • alastairp
        CatQuest: many people submitting
      • 2015-09-18 26109, 2015

      • CatQuest
        aha
      • 2015-09-18 26124, 2015

      • CatQuest
        I just... because I have like 18k pages as you know :P
      • 2015-09-18 26138, 2015

      • alastairp
        russss: we seem to have a problem when restarting wsgi doesn’t kill old child processes, I don’t think think that would be related though
      • 2015-09-18 26114, 2015

      • russss
        I enabled the Werkzeug profiler, which is saying something like this
      • 2015-09-18 26117, 2015

      • russss
      • 2015-09-18 26135, 2015

      • russss
        which is not *terribly* helpful as it doesn't tell you which socket
      • 2015-09-18 26142, 2015

      • russss
        oh wait hang on no
      • 2015-09-18 26154, 2015

      • russss
        urllib2 is your answer
      • 2015-09-18 26104, 2015

      • zxdert joined the channel
      • 2015-09-18 26151, 2015

      • alastairp
        we use urllib to connect to messybrainz
      • 2015-09-18 26114, 2015

      • russss
        if each call to messybrainz is taking 10ms and you've got 5000 calls to make, that would presumably do it
      • 2015-09-18 26122, 2015

      • russss
        I think messybrainz needs a bulk lookup API
      • 2015-09-18 26100, 2015

      • alastairp
        right
      • 2015-09-18 26133, 2015

      • russss
        that would probably be a sensible thing, I have a suspicion that messybrainz may be stalling on something as well so I'll look at that
      • 2015-09-18 26154, 2015

      • alastairp
        I checked indexes this morning, and it’s been a lot happier since
      • 2015-09-18 26113, 2015

      • alastairp
        I wonder if for the scale of listenbrainz we should bypass http when connecting to messybrainz
      • 2015-09-18 26125, 2015

      • russss
        might be sensible
      • 2015-09-18 26128, 2015

      • alastairp
        abstract it into a library that connects to the database directly
      • 2015-09-18 26132, 2015

      • alastairp
        since they
      • 2015-09-18 26140, 2015

      • alastairp
        ‘re both on the same machine for now
      • 2015-09-18 26131, 2015

      • russss
        uwsgitop localhost:5051 is the messybrainz uwsgi
      • 2015-09-18 26136, 2015

      • russss
        and it does seem pretty busy as well
      • 2015-09-18 26138, 2015

      • alastairp
        it would be more than LB actually, since it responds to 50 queries per single LB query
      • 2015-09-18 26157, 2015

      • zxdert has left the channel