#metabrainz

I've made that clear before and it is still true.

Leo_Verto: Ah, they're working.

Not sure why it was loading older page 🤔

too bad the summit isn't about you personally then ¯\_(ツ)_/¯

thre are more people than you.

Freso: can you please step in?

and fyi: YOU have not made my life that great either.

LordSputnik: Should I let the edition mapping for the time be, and work on wrapping up of works and creators import?

that's the last I wil lsay here

as I said. I am calm and patient

but even i can have my patience run out.

bukwurm: I'd go for that, yes

Wait to do editions later

LordSputnik: Ok

I'd like to see importing working this week

LordSputnik: Ok

It would be good to spend an hour or two with you tomorrow going through PRs and getting them closed up

I'll commit the creator and works mapping in an hour or so.

LordSputnik: Sure thing 👍

Ok, thank you

Is normal meeting time ok?

Yes

Ok good

LordSputnik: I had updated the SQL PR, but kept the last updated field on source as nullable

Anything you want to quickly chat about now?

As I felt that it's not necessary to always have a last_edited data

Is it ok?

Or are you happy to wait until tomorrow?

LordSputnik: Just the SQL PR doubt :)

bukwurm: ok - might be useful to document what will happen when stuff gets imported again with the same origin ID and there's no last updated in the db

I've changed it to timestamp, as was decided

Assume it needs reimporting?

LordSputnik: Yea, that's what I had in mind

Ok, that sounds fine to me

Did we agree anything about the timestamp timezone in the end?

LordSputnik: Yeah

I had a doubt what if the timestamp had a different timezone specified

But then realised we can change via some library and then check the timestamps

LordSputnik: I changed it to timestamp that same hour after the review :)

Ok I think I saw

LordSputnik: Ok, so I think I'll make a PR of creators and works mapping today

That would be good, thank you

And consumer side validators for them tomorrow

Ok, so by the time we meet tomorrow we should be able to import stuff?

LordSputnik: It'll still need the bb-data connection wrapping up

I'll try to finish it too, but can't be sure

We'll be:

Read, parse and map, push data into queue

Consume, validate the data, reject/accept

The only part left would be to push it to the db

I have basic setup done for the pushing to the db part, so hopefully it won't take much time in any manner

bukwurm: ok, try to get as much done as you can

LordSputnik: 👍

zas: ping

morning bitmap

moin zas

about rabbitmq, I ran into an issue setting up the consul peer discovery. for some reason the consul health checks don't work

so perhaps we shoul skip the cluster stuff for now

what did you use for health checks ? defaults ?

the checks are set up by the plugin (not using registrator)

15s ttl

ok, we can skip the cluster for now, we have to move everything before tomorrow 16:00 CEST

okay

9:00 AM for you i think

yup

there's also sentry on serge, who deployed it ?

i think we can stop it, and backup bamed volumes

sure, it's not super critical

serge crashed again few hours ago

and cpu temp was super high, but we go for a full replacement to be sure

I guess backup /root/.ssh/sentry_deploy* and /root/sentry-onpremise to be sage

safe

what does it need to run ?

i'm thinking about deploying on a VM

i think it's all contained in the compose file it provides

the ssh key seems to just be for pulling from our private sentry-onpremise repo (for upgrades)

SENTRY_POSTGRES_HOST: "10.2.2.24" <--

it uses bowie ?

oh yeah, it does

but for its own db ? or it needs other stuff ?

just its own db

we talked about moving it before

we can now i think, using a dockerized postgres server for it, i don't know how much resources it needs

but basically if we can spawn a VM and move everything sentry needs on it it'd be great

it doesn't seem integrate in our usual stuff

hmmm it uses exim4 container as smtp relay it seems, this will require a change in gmail config

yeah I guess it's based on the official sentry compose files

I'll put the new rabbitmq on trille unless you'd rather somewhere else

but apps are finding sentry, how do they?

trille is ok, that's temporary

we forked their stuff and modified the compose file to add SERVICE_* tags

can we afford to have sentry down few hours?

that'd be fine

yvanzo: ping

pong

¿Leo_Verto ^ Leo__Verto?

Matrix has a bit of lag, I'll use this

bitmap: for rabbitmq, you'll backup/restore rabbitmq-data, right? i think we can just do the same with sentry, we can stop caa-indexer and search-indexer

sshd-musicbrainz-fullexport <--- this one?

I was gonna try using the shovel plugin to move the data, but if we can stop the indexers, that'd be faster

we can stop them

I think sshd-musicbrainz-fullexport is unused, that was from when serge hosted ftp iirc

it isn't in https://github.com/metabrainz/docker-server-con...

i guess we can stop it right now

listenbrainz is also using rabbitmq, not sure what for

it does, that's why ruaok was whining ;)

i guess we don't use this rsyncd container either

gateways-redis isn't used atm (it is a failover)

not sure moving the volume will work very smoothly, the hostname is hardcoded in a ton of places

meant to link to the start of the thread

hmmm

unless we set the container hostname on trille to the old container id on serge? dunno

when you say hardcoded in many places, where ?

ruaok: ping

the directory names, but also in a bunch of binary files I think. I guess there are command to change it, but it seems tricky

i don't understand, this hostname is what atm?

it's the hostname inside the container, which defaults to the container id

on serge it's 053a018a9e71

oh, ok. And rabbitmq depends on it ?

yeah, it won't find the data if we just move it

that's stupid ;) but i think we can set this "hostname" on docker command line

--hostname , -h

in fact, if rabbitmq depends on this, it should have been done before

using a generic name (rmq-1), instead of letting docker generates a random one

but you can just start the new container with -h 053a018a9e71

not fancy, but should work

I'll try

zas: it's moved now, I think