#musicbrainz-devel

/

      • ruaok
        ok.
      • 2011-05-17 13743, 2011

      • warp
        let me know when I should start failing things again.
      • 2011-05-17 13703, 2011

      • ruaok
        get ready to do it. I'll start here.
      • 2011-05-17 13751, 2011

      • warp
        ok, I'm ready to submit for POST requests
      • 2011-05-17 13754, 2011

      • warp
        s/for/four/
      • 2011-05-17 13736, 2011

      • ruaok
        go!
      • 2011-05-17 13758, 2011

      • ruaok
        saw it.
      • 2011-05-17 13759, 2011

      • ruaok
        reverting
      • 2011-05-17 13701, 2011

      • warp
        yeah, now I get the traces in the browser.
      • 2011-05-17 13721, 2011

      • warp
        hm, body_parameters => {} in the Request object is complete, so all the data was submitted correctly by the browser.
      • 2011-05-17 13700, 2011

      • murdos
        ocharles: if you're still there, any idea about the reason of http://pastie.org/1918213 on my dev server?
      • 2011-05-17 13709, 2011

      • ruaok
      • 2011-05-17 13729, 2011

      • warp
        ruaok: seems cut off.
      • 2011-05-17 13732, 2011

      • ocharles
        murdos: sounds like some of your moose modules are out date
      • 2011-05-17 13741, 2011

      • ijabz_ joined the channel
      • 2011-05-17 13744, 2011

      • ocharles
        murdos: try and update MooseX::Role::WithOverloading
      • 2011-05-17 13747, 2011

      • ruaok
        warp: check your mail.
      • 2011-05-17 13752, 2011

      • ruaok
        I got email from catalyst. :)
      • 2011-05-17 13714, 2011

      • warp
        oh wow, with a little ascii error pointing at the offending line of source code. how fancy.
      • 2011-05-17 13720, 2011

      • warp
        ascii arrow.
      • 2011-05-17 13721, 2011

      • ruaok
        yeah. ;)
      • 2011-05-17 13736, 2011

      • ruaok
        lets use that more for fixing crash bugs
      • 2011-05-17 13756, 2011

      • ocharles
        it was meant to send those emails without CATALYST_DEBUG { 1 }
      • 2011-05-17 13710, 2011

      • warp
        ruaok: is there any limit to user sessions?
      • 2011-05-17 13713, 2011

      • murdos
        ocharles: perfect! it's working now! :)
      • 2011-05-17 13716, 2011

      • ocharles
        I thought my inbox was suspiciously empty
      • 2011-05-17 13746, 2011

      • ruaok
        warp: that question is best posed to ocharles
      • 2011-05-17 13754, 2011

      • ocharles
        "limit"?
      • 2011-05-17 13755, 2011

      • ruaok
        the sessions are stored in memcached.
      • 2011-05-17 13706, 2011

      • ruaok
        I wonder if something is falling out or getting clobbered.
      • 2011-05-17 13709, 2011

      • warp
        ocharles: is there a sizelimit to the sessions store for a user?
      • 2011-05-17 13728, 2011

      • ocharles
        maybe the memcached instance isn't big enough
      • 2011-05-17 13741, 2011

      • warp
        ocharles: because it looks like things are empty when the wizard tries to restore the values for a page from the serialized store (i.e. the session store).
      • 2011-05-17 13726, 2011

      • ruaok
        ocharles: thats easy to test.
      • 2011-05-17 13729, 2011

      • ruaok
        should I up it?
      • 2011-05-17 13738, 2011

      • ruaok
        that means everyone will be logged out. heh.
      • 2011-05-17 13738, 2011

      • ocharles
        that'd be my first line of action, yea
      • 2011-05-17 13741, 2011

      • ruaok
        ok
      • 2011-05-17 13749, 2011

      • ocharles
        warp: i don't think we're deleting session data on wizard submit either
      • 2011-05-17 13752, 2011

      • warp
        except for tracklists and recording associations everything in the release editor goes into the session. tracklists and recording associations only if they have been edited, then they also go in there.
      • 2011-05-17 13757, 2011

      • ocharles
        so memcached is going to fill up very fast
      • 2011-05-17 13707, 2011

      • ocharles
        i don't know how memcached decides what to drop
      • 2011-05-17 13712, 2011

      • warp
        ocharles: hehe, good point :)
      • 2011-05-17 13717, 2011

      • ocharles
        fifo?
      • 2011-05-17 13723, 2011

      • luks
        LRU
      • 2011-05-17 13751, 2011

      • warp would expect LRU
      • 2011-05-17 13755, 2011

      • ocharles
        luks: thanks, that's intuitive
      • 2011-05-17 13704, 2011

      • ocharles
        so, i'm not sure why only 4 sessions fit in
      • 2011-05-17 13712, 2011

      • ocharles
        but the session store must just be too small
      • 2011-05-17 13757, 2011

      • ruaok
        32mb right now.
      • 2011-05-17 13759, 2011

      • ruaok
        I'll double it
      • 2011-05-17 13713, 2011

      • warp
        that sounds absolutely tiny.
      • 2011-05-17 13730, 2011

      • ruaok
        restarted.
      • 2011-05-17 13741, 2011

      • ruaok
        its starting to. :)
      • 2011-05-17 13750, 2011

      • ruaok
        but we had not idea where to start.
      • 2011-05-17 13709, 2011

      • warp
        ok, I'll go set up those four POSTS again.
      • 2011-05-17 13741, 2011

      • ruaok
        the four POSTmen of the apocalypse?
      • 2011-05-17 13756, 2011

      • warp
        got four Internal Server Errors now.
      • 2011-05-17 13757, 2011

      • warp
        :(
      • 2011-05-17 13715, 2011

      • ruaok
        did you log back in?
      • 2011-05-17 13722, 2011

      • warp
        no
      • 2011-05-17 13735, 2011

      • warp
        I'm still logged in.
      • 2011-05-17 13747, 2011

      • ruaok
        odd.
      • 2011-05-17 13752, 2011

      • ruaok
        maybe log out and back in?
      • 2011-05-17 13706, 2011

      • warp
        ok
      • 2011-05-17 13724, 2011

      • warp
        I do have "Keep me logged in" checked, so it isn't that odd.
      • 2011-05-17 13734, 2011

      • ocharles
        yea, that actually works now
      • 2011-05-17 13745, 2011

      • ocharles shakes a fist at unicode
      • 2011-05-17 13747, 2011

      • ruaok
        I wonder if it can't rebuild the session on a post?
      • 2011-05-17 13719, 2011

      • luks
        getting 500 on login now
      • 2011-05-17 13734, 2011

      • luks
        it was also telling my I entered incorrect password before
      • 2011-05-17 13750, 2011

      • warp
        ruaok: two internal server errors. two of them completed.
      • 2011-05-17 13706, 2011

      • ruaok
        thats better, right?
      • 2011-05-17 13718, 2011

      • warp
        ruaok: I've had that happen before.
      • 2011-05-17 13751, 2011

      • ruaok
        luks: is that persisting? I can log in.
      • 2011-05-17 13701, 2011

      • ruaok
        with with permanent and without
      • 2011-05-17 13724, 2011

      • luks
        ruaok: no
      • 2011-05-17 13739, 2011

      • ruaok
        odd
      • 2011-05-17 13752, 2011

      • ruaok chalks it up as an artifact of restarting the session store
      • 2011-05-17 13711, 2011

      • warp
        ruaok: when printed out with Data::Dumper two of these sessions are 114kbyte.
      • 2011-05-17 13720, 2011

      • warp
        ruaok: so that really should fit in 64mb :)
      • 2011-05-17 13730, 2011

      • ruaok
        yeah
      • 2011-05-17 13714, 2011

      • warp
        I wonder if we're using the catalyst session store correctly with regards to having multiple tabs open.
      • 2011-05-17 13742, 2011

      • ocharles
        what do you think we could be using wrong?
      • 2011-05-17 13755, 2011

      • ruaok
        that we're clobbering the other sessions in the store
      • 2011-05-17 13707, 2011

      • ocharles
        but all wizard sessions are keyed on a unique random number
      • 2011-05-17 13718, 2011

      • warp
        ocharles: well, I cannot otherwise explain why data is going missing. perhaps the POST submit of subsequent tabs somehow overwrites the existing session.
      • 2011-05-17 13741, 2011

      • ocharles
        warp: are you able to reproduce locally, with 2 REs going on at once?
      • 2011-05-17 13750, 2011

      • warp
        ocharles: haven't tried, let me do that now.
      • 2011-05-17 13731, 2011

      • warp
        ocharles: they all succeed here.
      • 2011-05-17 13740, 2011

      • ocharles
        hrm
      • 2011-05-17 13743, 2011

      • warp
        (using script/musicbrainz_server.pl -r)
      • 2011-05-17 13702, 2011

      • warp
        sub SESSION_STORE { "Session::Store::File" }
      • 2011-05-17 13705, 2011

      • ocharles
        right
      • 2011-05-17 13720, 2011

      • luks
        what's the difference between NGS-Final and NGS-CI?
      • 2011-05-17 13732, 2011

      • luks
        http://tickets.musicbrainz.org/browse/MBS-1916 should be probably fixed soon, it makes voting on some release changes impossible
      • 2011-05-17 13742, 2011

      • ocharles
        luks: nothing really, because they are both going to release at the same time
      • 2011-05-17 13749, 2011

      • ocharles
        atm we're just treating CI as "super important"
      • 2011-05-17 13759, 2011

      • ocharles
        luks: i'm working on those atm
      • 2011-05-17 13705, 2011

      • ocharles
      • 2011-05-17 13705, 2011

      • luks
        thanks
      • 2011-05-17 13711, 2011

      • warp
        ok, it's getting too late for me.
      • 2011-05-17 13722, 2011

      • murdos found a funny bug: http://tickets.musicbrainz.org/browse/MBS-1994
      • 2011-05-17 13735, 2011

      • ocharles
        heh, nice
      • 2011-05-17 13705, 2011

      • luks
        does somebody remember how the old code handled situations like this http://ngs.musicbrainz.org/edit/14461980 ?
      • 2011-05-17 13713, 2011

      • luks
        removing an AR from a release that was merged
      • 2011-05-17 13726, 2011

      • luks
        so the AR is now attached to another release
      • 2011-05-17 13754, 2011

      • ocharles
        fuck. all release group edits before http://ngs.musicbrainz.org/edit/14449260 are corrupt
      • 2011-05-17 13701, 2011

      • ocharles
        (ngs ones, that is)
      • 2011-05-17 13727, 2011

      • ruaok
        how many are there?
      • 2011-05-17 13733, 2011

      • ocharles
        scrap that, the historic ones are corrupt
      • 2011-05-17 13742, 2011

      • ocharles
        they use an old artist credit definition
      • 2011-05-17 13745, 2011

      • ruaok
        even joyier.
      • 2011-05-17 13754, 2011

      • ocharles
        the data is still correct, but we need to change the schema
      • 2011-05-17 13701, 2011

      • warp
        ruaok: I've briefly documented what we know about the session issue so far on http://tickets.musicbrainz.org/browse/MBS-1952 .
      • 2011-05-17 13720, 2011

      • ruaok
        ok, crack on in the morning with that, plz
      • 2011-05-17 13723, 2011

      • ruaok
        ocharles: what change?
      • 2011-05-17 13731, 2011

      • warp
        ruaok: well, I cannot reproduce it here.
      • 2011-05-17 13746, 2011

      • ocharles
        warp: historic = "artist_credit":[{"artist":"525479","name":"Николaй Гиршевич Кaпустин"}]
      • 2011-05-17 13702, 2011

      • ocharles
        sorry, that's to ruaok
      • 2011-05-17 13706, 2011

      • ocharles
        ngs = "artist_credit":{"names":[{"join_phrase":null,"artist":{"name":"Николaй Гиршевич Кaпустин","id":"525479"},"name":"Николaй Гиршевич Кaпустин"}]}
      • 2011-05-17 13709, 2011

      • warp
        ruaok: I can try to replicate the production setup I guess, either on hobbes or on my local machine. can you send me a production DBDefs?
      • 2011-05-17 13722, 2011

      • ruaok
        sure
      • 2011-05-17 13722, 2011

      • ocharles
        ruaok: as http://tickets.musicbrainz.org/browse/MBS-1960 is a historic edit problem that can be easily fixed, i'm moving it to NGS+1
      • 2011-05-17 13747, 2011

      • ruaok
        ocharles: is it a schema change to the edit stuff?
      • 2011-05-17 13704, 2011

      • ocharles
        ruaok: it means I have to UPDATE all add release group edits that are historic
      • 2011-05-17 13729, 2011

      • ruaok
        warp: mailed.
      • 2011-05-17 13737, 2011

      • ocharles
        It's a pretty simple JSON transformation
      • 2011-05-17 13747, 2011

      • ruaok
        good. and its not on any replicated tables?
      • 2011-05-17 13715, 2011

      • warp
        ruaok: thanks, got it.
      • 2011-05-17 13723, 2011

      • ruaok
        if it doesn't touch replicated tables and we don't have to re-migration then GO
      • 2011-05-17 13738, 2011

      • ocharles
        ruaok: it just touches rawdata.musicbrainz.edit
      • 2011-05-17 13751, 2011

      • ruaok
        phew.
      • 2011-05-17 13752, 2011

      • ruaok
        awesome.
      • 2011-05-17 13755, 2011

      • warp
        ruaok: if there's anything else which needs my immediate attention tomorrow morning, put it in my current iteration in jira.
      • 2011-05-17 13704, 2011

      • ruaok
        will do. thx.
      • 2011-05-17 13710, 2011

      • ruaok
        tomorrow morning we should have a meeting.
      • 2011-05-17 13715, 2011

      • warp
        (or poke me on irc, or e-mail me.. perhaps I'll check e-mail :), etc... )
      • 2011-05-17 13723, 2011

      • ruaok
        and see if tomorrow evening 16:00 is plausible for lauch
      • 2011-05-17 13728, 2011

      • ruaok
        k
      • 2011-05-17 13732, 2011

      • ruaok
        *launch
      • 2011-05-17 13733, 2011

      • warp nods.
      • 2011-05-17 13704, 2011

      • ocharles
        16 where?
      • 2011-05-17 13730, 2011

      • ruaok
        BST
      • 2011-05-17 13734, 2011

      • ocharles
        ok
      • 2011-05-17 13751, 2011

      • ruaok
        just guessing for now.
      • 2011-05-17 13706, 2011

      • ruaok
        its seems the serious bugs are getting fixed.
      • 2011-05-17 13736, 2011

      • djce
        ruaok: When do you anticipate deciding on a launch time?
      • 2011-05-17 13709, 2011

      • ruaok
        I'd like to have a meeting tomorrow morning 8am BST time to decide.
      • 2011-05-17 13723, 2011

      • ocharles
        then I think I'll head off in the next 2 hours