#metabrainz

Leftmost: good idea starting that thread, will reply tomorrow :)

Anybody around that would be willing to help me connect to the musicbrainz API running on my local VM using musicbrainzngs? I keep getting a HTTP 503 error

what sort of queries are you doing? if they're search queries, you are probably still hitting musicbrainz.org unless you set up a search server

I'm hosting musicbrainz locally (within VM) and set musicbrainzngs.set_hostname("localhost:8080")

I can connect to localhost:5000 and pull HTML correctly. I feel like I'm missing something

to answer your question, I'm running musicbrainzngs.search_recordings

what is LUCENE_SERVER set to in lib/DBDefs.pm? should be localhost:8080

musicbrainzngs.musicbrainz.NetworkError: retried 8 times, caused by: None

LordSputnik, I'm full of good ideas. :)

If only I knew what any of them were.

and I think musicbrainzngs.set_hostname should be called with localhost:5000, not directly to the search server

since the query parameters are different if you're querying lucene directly

^ I've also tried localhost:5000

lib/DBDefs.pm is where? in the python module or on the VM guest OS?

PS. thanks for your help

in the VM. according to http://wiki.musicbrainz.org/MusicBrainz_Server/... it's at /home/musicbrainz/musicbrainz-server/lib/DBDefs.pm

it's set to localhost:8080

ok, good

I've also tried Robert's suggestion here: http://tickets.musicbrainz.org/browse/MBVM-3 (didn't get me any further)

not really sure what's up yet. can you access http://localhost:8080/ws/2/recording/?query=foo in your browser and does it return results?

Interesting, I got: curl: (56) Recv failure: Connection reset by peer

If I run the same thing on port 5000 I get: Could not retrieve sub-document page from search server. Error: http://localhost:8080/ws/2/recording/?max=25&am... -> 500 Index is currently not available for resource type RECORDING

oh. have you not built any search indexes yet?

no, I wasn't sure if the image shipped with search indices

I'll run it (it'll take a while I'm guessing)

nope, they're way too big to ship with the vm

that makes sense

cool, I think that's probably my issue

Gentlecat: The only need for batch processing arises when we need to show multiple reviews on a page (used by review view). This is presently handled by multithreading by get_multiple_entities(). So to do this job in batch I will not use multithreading anymore, instead query the database for all the entities at once by their ID and return the result obtained. Is this okay? Am I missing something? Thanks :D

something like that, yes

Thanks a lot :D

Morning all, had a couple more replication failures overnight. Both connection timeouts again.

> Congrats @omnifone and @c4dm for launch of Acoustic Lab! Can’t wait to see and compare data. How do we get access?

> @AcousticBrainz @c4dm Contact sales@omnifone.com for more information!

that was entirely predictable. :)

hah

awesome

well, that settles it.

chrisskye: can you please add omnifone to your hit list? they use our data.

ok.

see https://twitter.com/Omnifone/status/71257273007...

when someone one responds to us with a sales email, we can hit them for money.

alastairp: now that I see your nick, I just remembered a dream I just had.

you were president elect.

as a kiwi at that. it was weird, but then again dreams nearly always are. :)

So they use our data for free, but want to sell to us?

wait

what?

chirlu: YES. Make sure to let them know their fuckup.

sorry chirlu, that was meant for chrisskye

alastairp: exactly. :)

gburlet: hello?

alastairp: my todo list is getting quiet and I am hoping to use semana santa as an excuse to power though a bunch of LB stuff.

I know you're buys this week, but when you have a moment for a couple of questions about your recent discussions, I'd appreciate it.

ruaok: I can't believe they said that, and wasted our time with their inconsistent support :(

so, there is a bit a of lesson to be learned from Spain on this.

try again. be super sweet, chat them up.

make them realize the hardship you're enduring and somehow compel them to do their jobs.

a public appeal might also work.

ruaok: I was thinking that, but that would be the fourth ticket and they might not even read it :P

I've emailed navap to see if he's received a response

It *should* be simple for him to get it back, since he has control of the registered email

yeah.

hey navap, help us out here, please?

ruaok: yeah, I can talk

cool.

the cassandra -> PG thing is clear in my head.

I'll post a WIP schema review for that.

the rate limiting depends on what we do with the last.fm API scraping timestamp resolution issue.

and the timestamp issue is what I am most unclear on.

did you suggest that we should call the API from the browser (can we do that?) or that we should make import work on the server side?

then add a queue for importing profiles and just keep working on the queue?

yes, we can call the api from the browser

CORS allowed and all that?

to me it's a pretty reasonable thing to do - especially since it means that we don't become a bottleneck

apparently so

armalcolite's patch works

ok, good. that is sufficiently decoupled from the PG work.

has that been merged yet?

no, we did some architechture rework, but I still need to revise his changes

or someone

I can have a look.

I think I will have a look at that and also see about nailing the rate limiting BS finally.

then I can worry about the PG backend and other cleanup. then we focus on a new release and considering if we wipe the DB.

chrisskye: very good email. smack 'em hard!

> As such, this tweet with the suggestion to “contact sales” was taken as a bit offensive by our team:

\ø/

thanks. :)

These emails aren’t as easy to write as they may appear...

I know what you mean.

fine line on a lot of topics.