#metabrainz

mo’’in’

[CAA-indexer] yvanzo merged pull request #16 (master…rabbitmq-service-name): Support RABBITMQ_SERVICE environment variable https://github.com/metabrainz/CAA-indexer/pull/16

good morning!

Mornin !

What is correct? "take down", "take-down" or "takedown" request?

alastairp: that seems like the kind of thing you'd have an opinion about for some reason

(I just saw we have it as "take down" in one place and one of my PRs adds "takedown" in another...)

ruaok: you too ^ :)

reosarevok: https://www.dmca.com/faq/What-is-a-DMCA-Takedown

mooooin!

Ok, then we should fix the one we have. Dammit :D

(and thanks)

[musicbrainz-server] reosarevok opened pull request #1917 (master…MBS-11389): MBS-11389: Top-align edit medium Tracklist tables https://github.com/metabrainz/musicbrainz-serve...

Project bookbrainz-site build #3649: passed in 4 min 33 sec: https://travis-ci.org/bookbrainz/bookbrainz-sit...

Project bookbrainz-site build #3651: passed in 4 min 26 sec: https://travis-ci.org/bookbrainz/bookbrainz-sit...

ruaok: alastairp: i am trying to fix the failing test here https://ci.metabrainz.org/job/listenbrainz-spar.... It seems to happen because the test mocks subprocess.Popen which is also patched during sentry initialization. I have guarded the sentry_initialization so that it only happens in case there is a `LOG_SENTRY` variable in `config`. However, it still fails. How is the config set during spark tests?

[musicbrainz-server] reosarevok opened pull request #1918 (master…MBS-11395): MBS-11395: Update the sidebar matching for muziekweb https://github.com/metabrainz/musicbrainz-serve...

thanks alastairp! it was using the sample config for tests which had an empty LOG_SENTRY variable

[musicbrainz-server] reosarevok opened pull request #1919 (master…MBS-11397): MBS-11397: Show place-series rels on series page https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] reosarevok opened pull request #1920 (master…MBS-11396): MBS-11396: Show item_entity_type on Series/CollectionType admin tables https://github.com/metabrainz/musicbrainz-serve...

yvanzo: are we ready for trille reboot or is the rabbitmq switch to prince incomplete yet?

ruaok: all tests passing now in spark sentry pr.

[listenbrainz-server] amCap1712 opened pull request #1290 (master…master): Use BU ratelimiting module https://github.com/metabrainz/listenbrainz-serv...

alastairp: thoughts on BU-27? the current workaround seems reasonable to me.

BU-27: allow disabling ratelimiting during automated testing of api endpoints https://tickets.metabrainz.org/browse/BU-27

zas: I’m building a new caa image now, last missing piece to migrate /caa to rabbitmq-prince

There still is one lb connection to rabbitmq-trille.

zas: have you moved the LB RMQ stats over to the next RMQ instance?

_lucifer: finishing the RMQ migration is key before we can test the spark stuff

ruaok: which dashboard is it?

we still have one connection to the old RMQ instance that we can't resolve which it is. we think it might be telegraf.

Migrating CAA-indexer to use RabbitMQ on prince now.

yes, likely, but this telegraf instance will notice when the service is stopped, and reconfigure

it runs on yehudi

yvanzo: ok, ping me when you're done with the CAA mirgration and we can rename the official service.

zas: see the last line. how can we tell what the real IP is that is making this connection?

can you know which user it uses? telegraf uses guest

(about real IP... well, docker ...)

this connection is using guest, yes.

[musicbrainz-server] reosarevok opened pull request #1921 (master…MBS-10195): MBS-10195: Block Google AMP links https://github.com/metabrainz/musicbrainz-serve...

ruaok: done :)

yay.

has the consul service been updated so that we no longer need the ENV var hack to find the new rabbitmq?

bitmap: I reverted your commit to work around this error: https://gist.github.com/yvanzo/d72622511ba9579d...

ruaok: only for CAA.

yvanzo: not sure I asked clearly enough. Has the RMQ service name been renamed so that a service (listenbrainz spark) that doesn't know how to connect to the temporary service will now connect to the new service?

ruaok: no, rabbitmq-prince is the new service.

ruaok: ListenBrainz images need RABBITMQ_SERVICE set to 'rabbitmq-prince' to connect to it.

yes, understood.

Replacing that env var to just use consul-template instead would need a patch to LB and rebuilding images again.

we have one service whose setup has not been patched at all, that is why i am asking.

There currently are 3 connections using 'caa' user and 1 using 'lb' user to the old 'rabbitmq' on 'trille'.

CAA-indexer is not possibly doing that, so maybe the 'lb' connection doesn't come from an LB container either.

oh, never mind. the spark request consumer is not part of the consul world, it needs to be manually updated in a much more complicated process. ok, I'll get that process going.

zas says that the telegraf configuration will automatically pick up the change an reconfigure once the rabbitmq service as defined by consul changes from rabbitmq to rabbitmq-prince.

like here: https://github.com/metabrainz/listenbrainz-serv...

will that service be renamed or has it already been renamed?

what would happen if I took out the latter half of line 64? would the service work now or break?

zas: ping. i need your help on something.

pooong

hey.

remember how trille has one firewall hole in it to allow RMQ traffic from leader.listenbrainz.org ?

Prince now hosts the new rabbitmq and this firewall hole needs to move from trille to prince? can you please do that for me?

s/prince?/prince./

the allowed IP should be 144.76.198.54

I didn't remember, but now yes ;) 65413/tcp ALLOW 144.76.198.54 # rabbitmq_leader.listenbrainz.org

that.

same port?

it needs to be on prince now.

yes, plz.

done

I'm almost done double checking that this is right. lol. :)

shit, sorry.

62673 is the correct port. :(

zas? ^^

ok :) fixing it

thx

shoot again

works, thanks!!

[musicbrainz-server] reosarevok opened pull request #1922 (master…MBS-5680): MBS-5680: Allow inc=artist-credits in combination with rels incs that have ACs https://github.com/metabrainz/musicbrainz-serve...

I just closed the 3 connections for 'caa' on old trille instance and they did not reopen.

However, the connection for 'lb' cannot be closed: it's still getting reopened.

yvanzo: as said; if it is telegraf it will just remove this instance from its config when the container will be stopped