ruaok or anyone who has access to MeB travis: please activate builds for LB
CatQuest
but not pay ot use *mb* itself proper
higher ws stuff
ruaok doesn't
iliekcomputers_
ruaok: Yeah I figured :)
SothoTalKer
have 2 ws, one free with limits, one not free and more 'unlimited' :)
ruaok
I'm busy trying to figure out how to work nicely with rabbitmq
CatQuest
anyway I go to showever. the fireplace is clean(ish), but I am dirty :D
SothoTalKer
if the api key is per developer and not per user, anyway...
CatQuest
rabbits!
iliekcomputers_
ruaok: what happened to the one connection per worker plan?
ruaok
trying that now.
reosarevok
CatQuest: so you were dealing with ash after all :D
Freso
SothoTalKer: There could also be OAuth2 authentication on top of the API key to do per-user stuff regardless of API key being used.
SothoTalKer
still, to run my script i would need to buy an api key from you :D
CatQuest
reosarevok: i was :D
iliekcomputers_
zas: would you mind activating build s for LB on Travis please? I see picard builds so I hope you have access.
exogen joined the channel
Freso
(E.g., so a user who is part of paying for the "less-crowded server" could log in with OAuth2 and get access to the limited server via beets and Picard and Kodi and whichever other apps they use that use MB.)
SothoTalKer: You... were not listening? API keys are free.
CatQuest presumes sotho is *still* beign sillyjoking
CatQuest regrets not buying more salt candy T__T
SothoTalKer
i was, just wanted to read it again
Freso: can i use my MB user credentials to access the ws API?
MatthewGlubb
Just reading through your comments about traffic concerns, I'm not familiar with your setup or traffic profile - e.g. cache hits, etc but couldn't you stick Varnish in front of your web service? It strikes me that most of your traffic must be GETs and therefore imminently cacheable. Varnish is very, very configurable, and can be clustered. You can even cache fragments of responses but this would require some work in the app layer. I'll
I can't remember for how long the conversation went on, but we literally just had it a few days ago.
Caching is definitely not ruled out, but it's not so easy as "just do X".
MatthewGlubb
No silver bullet to anything
yvanzo
MatthewGlubb: We have been recently looking at Varnish for the website.
kepstin
on the other hand, even if there's a very long tail, it might be that adding caching just to help the popular requests would be enough to free up more resources to handle the more unique requests.
Freso
SothoTalKer: Also, keep in mind that all this talk is for a future /ws/ version that has not seen a single line of actual code yet. And /ws/2 will be available for at least some years to come still. (Heck, /ws/1 is still available, after years of being deprecated.)
SothoTalKer
Freso true :)
Freso
Actual implementation with regard to rate limit, authentication (API keys, OAuth2), etc., etc. is still mostly entirely up in the air.
to81 joined the channel
reosarevok
ruaok: that latest NWR/CRI reply is very encouraging :) Definitely interested to check it all out now
ruaok
yeah, me too.
any thoughts on format?
reosarevok
Not sure, any idea what they have?
A JSON dump might work?
I'm honestly not sure of the volume of data we're talking about
ruaok
for a test, I think she said 1,200 releases.
I'll just ask them what they use for a DB and go from there.
reosarevok
Well that's what their label has :)
But yeah, let's do that
A DB dump would be interesting, if they just store JSON text that would honestly probably be useful too
As long as it's not excel...
SothoTalKer
i bet its excel
reosarevok
ruaok: I'm happy to have passed the mantle of local pessimistic dude to SothoTalKer
ruaok
if they send an excel file, they've already failed.
reosarevok: sigh.
as long as he doesn't show up at the summit all is well, right?
arbenina_ has quit
SothoTalKer
i don't plan to, don't worry :)
ruaok
one of you two at the summit is enough.
and thank FSM there is no oven at the office.
kyan joined the channel
kyan has quit
not excel. :)
kyan joined the channel
boneskull has quit
boneskull joined the channel
boneskull has quit
exogen has quit
antgel joined the channel
arbenina_ joined the channel
reosarevok
Definitely not excel
reosarevok is kinda excited
Freso too :)
Plus unlike NT this one is straight in the middle of my editing interests
So I am more than willing to work on it
samj1912 has quit
regagain has quit
Freso
NT?
reosarevok
Ninja Tune
Which I was supposed to look into, but eh :/ There's only that much excel a human can deal with
Freso
Ah.
kyan has quit
arbenina_ has quit
github joined the channel
github
[listenbrainz-server] mayhem opened pull request #190: Flask pika (master...flask-pika) https://git.io/vHwIR
github has left the channel
ruaok
travis builds are happening. thanks zas and iliekcomputers_
kyan joined the channel
iliekcomputers_
🎉
reosarevok
But Travis is just a shit Coldplay!
iliekcomputers_
And Coldplay is a shit Radiohead :)
reosarevok
They wish
Clint
and radiohead is a shit band ;p
reosarevok
Eh
There's worse stuff :p
iliekcomputers_
Like muse ;)
Clint
like coldplay
SothoTalKer
möp
luks
is reimport the only way to fix a replicated db? and if so, it the 20170605-031203 data dump ok?
ruaok
there was a hiccup today that bitmap is working on. if you hit that problem, might be good to wait.
luks
ok, I'll wait then
bitmap
I need to test the new packet first, might take an hour or two. but yes, that data dump is ok
there is an actual bug in our alias merge code which confuses dbmirror, that won't be fixed until later, but I guess it's unlikely it'll cause any issues again if it took this long
luks
thanks, I'll reimport my dbs tomorrow
regagain joined the channel
bitmap
luks: when you said you got the same error with mbslave earlier, were you talking about this artist_alias issue, or the "value too long for type character varying" encoding issue?
luks
artist_alias
bitmap
okay
ferbncode_ has quit
then yeah, the packet is wrong... the DBD::Pg stuff was for something else