monkey: i think mayhem will need to check spotify dashboard and update the uris there if needed. iirc test.lb is using a separate spotify app from beta and prod.
2022-10-31 30413, 2022
mayhem
one sec
2022-10-31 30415, 2022
monkey
Yep, was just writing that I don't have access to the spotify account
ah, interesting. I guess we're using the same app on both prod and dev?
2022-10-31 30433, 2022
monkey
thank you*
2022-10-31 30446, 2022
mayhem
no, but we could.
2022-10-31 30413, 2022
mayhem
updated, monkey
2022-10-31 30434, 2022
monkey
Thank you
2022-10-31 30447, 2022
alastairp
maybe remove the test callback urls for the prod instance? to reduce possible future confusion
2022-10-31 30454, 2022
lucifer
monkey: you'll have to write spotify gdpr support that you want timestamps of the export in milliseconds. if they still send in seconds then write back to them emphasizing that its seconds but you want in milliseconds.
2022-10-31 30456, 2022
mayhem
works now.
2022-10-31 30400, 2022
mayhem
alastairp: agreed.
2022-10-31 30423, 2022
mayhem
done
2022-10-31 30424, 2022
monkey
OK, will try, thanks lucifer.
2022-10-31 30427, 2022
alastairp
great
2022-10-31 30430, 2022
lucifer
alastairp: yeah, it had taken some back and forth but i was able to get ms precision timestamps last time.
2022-10-31 30437, 2022
alastairp
lucifer: oh! wow!
2022-10-31 30446, 2022
alastairp
and you're not even from europe :)
2022-10-31 30403, 2022
monkey
Alternatively, I might just sort the playback history and try to import all the ones that are before my first LB listen
oh i sent this first email to privacy@spotify.com not support. rest of comms were with support.
2022-10-31 30411, 2022
chinmay
alastairp: using linux, yeah
2022-10-31 30451, 2022
chinmay
lucifer: monkey: removing listenbrainz/webserver/static/dist worked
2022-10-31 30419, 2022
outsidecontext
took me quite a while to figure out that Picard Windows code signing was failing because MBs US address has changed :( Assumed it is me too stupid to use openssl properly
2022-10-31 30439, 2022
outsidecontext
hurray for cryptical error codes instead of human readable messages
2022-10-31 30448, 2022
mayhem
ohhh, that sucks.
2022-10-31 30407, 2022
mayhem
and on top of that, our address has to change again, since our mail forwarding company moved. 🤦
2022-10-31 30455, 2022
outsidecontext
ha :) but the certificate is done now and valid for a while. the problem here was just that for the windows app package you need to set the publisher, and that must match the certificates subject. but publisher was hard coded
2022-10-31 30401, 2022
mayhem
monkey: played 8 tracks without stopping!
2022-10-31 30416, 2022
monkey
Gosh, the bar really is low !
2022-10-31 30431, 2022
monkey
And I am definitely seeing working token refreshing, so I think that's that particular problem solved ! W00t!
Looks like it's not just me being a little slow with the spotify SDK token refresh thing. Their own example application gives bad advice as to how to set the player up: https://github.com/spotify/spotify-web-playback-s…
2022-10-31 30432, 2022
mayhem
damnit. player stopped. network tab not open.
2022-10-31 30414, 2022
mayhem
lucifer: not for LB#2226, but a later one.... I think for the similarity stuff we should exclude the seed artist_credit from the similar recordings
this commit implements an exclusion so that people who try to make an API request off listenbrainz.org (and not api.listenbrainz.org) are given the error message and the 410 error.
correct. once we are ready to deploy this PR we will move it to gateways.
2022-10-31 30432, 2022
zas
it would be easier to do the reverse, because I need names (like prod.listenbrainz-web), the rest is basically a copy'n'paste. Do you know which target names it will use on prod?
2022-10-31 30457, 2022
mayhem
kiss, I would expect.
2022-10-31 30421, 2022
zas
yes, but which container?
2022-10-31 30422, 2022
mayhem
should we do this after we deploy this PR and move the DNS entry?
2022-10-31 30439, 2022
mayhem
one sec.
2022-10-31 30446, 2022
zas
it will be easier for sure, as is it would be hard to test
2022-10-31 30401, 2022
mayhem
ok, I'll come back to you when we deploy this.
2022-10-31 30401, 2022
zas
(since files will not be used until target containers are up)
2022-10-31 30421, 2022
mayhem
but it will be listenbrainz-art-{test|beta|prod}
2022-10-31 30422, 2022
zas
the rest is basically a copy'n'paste of what we have for LB
2022-10-31 30426, 2022
mayhem
ok, cool.
2022-10-31 30427, 2022
mayhem
thanks!
2022-10-31 30434, 2022
zas
I'd recommend to stick to (prod|beta|test).service
2022-10-31 30402, 2022
zas
maybe prod.listenbrainz-art ?
2022-10-31 30422, 2022
mayhem
yes, since what the gateway uses is slightly different than the actual container names.
mayhem: i guess we can add a flag for that filtering? or should we do it always
2022-10-31 30458, 2022
yvanzo
zas: on it
2022-10-31 30415, 2022
yvanzo
zas: all good
2022-10-31 30401, 2022
lucifer
yvanzo: thanks for the logs, i'll look into it. btw is this a blocker for doing a new releaes?
2022-10-31 30457, 2022
yvanzo
lucifer: yes
2022-10-31 30424, 2022
lucifer
ah ok 👍
2022-10-31 30452, 2022
lucifer
yvanzo: from a quick search of the db error message, it seems benign. and debug logs of sir do not seem to indicate any issues either? did it crash or completed normally?
I mean it doesn’t return but events seem to be all processed.
2022-10-31 30450, 2022
lucifer
hmm, weird indeed. and it only happens for events?
2022-10-31 30409, 2022
yvanzo
no, I mentioned it can happen for some others.
2022-10-31 30424, 2022
yvanzo
events, release groups, urls...
2022-10-31 30429, 2022
lucifer
oh ok.
2022-10-31 30421, 2022
yvanzo
it seems that Postgres log lines more or less match loading relationship in SIR debug.
2022-10-31 30433, 2022
lucifer
the PG log lines seem unrelated to me. my intuition is that some thing related to multi processing is hanging up but could be wrong. i'll try to reproduce the issue and see if i can fix it.
2022-10-31 30454, 2022
CatQuest
..
2022-10-31 30442, 2022
mayhem
lucifer: do it always. I can't see how that data is any good.
2022-10-31 30423, 2022
lucifer
i see. we could also try doing it on spark level then?
2022-10-31 30429, 2022
mayhem
spark level would be best.
2022-10-31 30445, 2022
lucifer
cool, will do
2022-10-31 30447, 2022
mayhem
there is no need to move those lines around, they are useless.
2022-10-31 30436, 2022
mayhem
and can you please run the following 4 days sets for me? d 1095, s 100, t 5; d 730, s 100, t 5; d 1095, s 200, t 5; d 730, s 200, t 5;
2022-10-31 30453, 2022
lucifer
sure
2022-10-31 30404, 2022
mayhem
th
2022-10-31 30406, 2022
mayhem
x
2022-10-31 30434, 2022
mayhem
time to see what the session does. I keep getting 2-3 tracks of heavy metal in this test playlists. when it doesn't belong.
2022-10-31 30433, 2022
darkstardevx has quit
2022-10-31 30447, 2022
darkstardevx joined the channel
2022-10-31 30411, 2022
yvanzo
zas: It seems that SIR is 10x slower than usually but I cannot explain why atm.
2022-10-31 30426, 2022
yvanzo
(Restarting it didn’t help either.)
2022-10-31 30448, 2022
atj
just 10x?
2022-10-31 30456, 2022
yvanzo
atj: Actually more 8x or 9x but I round it up.
2022-10-31 30414, 2022
atj
when did this start, do you know?
2022-10-31 30420, 2022
atj
roughly
2022-10-31 30459, 2022
yvanzo
atj: 14:50
2022-10-31 30418, 2022
yvanzo
+-2min
2022-10-31 30409, 2022
CatQuest
mayhem: heavy metal always belong
2022-10-31 30410, 2022
Freso
Oh, btw. Meeting is in ~7 minutes for anyone who doesn’t follow EU DST. bitmap, akshaaatt, lucifer, CatQuest, Pratha-Fish, Shubh, yellowhatpro, ansh, riksucks, … uh… and anyone else I may have missed.
