#metabrainz

right, it uses the build/image trick

if you have both in a docker-compose file, it performs the build and then tags it with `image:`

oy. I think some tuning on ts is necessary.

it drives gaga to 98% disk space twice a day.

oops. anything in logs about what it's doing?

recreating those materialised views?

I'm guessing that is it.

I'll have to have a read on how to tune that.

ah ok! makes sense.

_lucifer: I just added ./annoy_indices to dockerignore

I told you it was hacky :)

a lot of our work is :D

oh - the other hacky part is that we're using postgres running directly on bono, and mounting in /var/run/postgresql to the image so that it can connect

in other news, I just tested a build and BU + setuptools dependency with no versions makes it try and install a version which isn't python-2 compatible

yup, i guessed that was the case seeing the volumes.

in BU 2 or 1.18?

weirdly it's using 1.15

oh i see i changed the branch to download_ds.

i should probably merge master into the branch first.

it's possible it's due to https://github.com/metabrainz/brainzutils-pytho...

so we should be able to fix it by explicitly installing a lower version of that before installing from requirements.txt

(the discussion we had about the model training PR will allow us to upgrade to py 3)

I don't know the answer to either of those questions :)

we updated AB to 1.18.1

but because the dependency for setuptools_scm in BU's setup.py has no version number, it seems like they released a new version which dropped py 2 support

oh!

makes sense.

we could release BU 1.18.2 which includes a version restriction for AB

sure, let's do that.

OK, let me do that. we discussed that we'll need a branch if we continue to make 1.18 releases. what do you want to call it? 1.18, or 1.x ?

there's already one iirc.

v1.18

oops, missed the fact that we made one of those. thanks

although - now I'm a bit confused. latest setuptools_scm was released March 17, but our change was made March 31

maybe it's a dependency of a dependency

yeah, just saw that.

they dropped python2 support in 6.0.0.

maybe on March 31, somehow we still got the 5.* version so it worked?

actually, its building fine now.

yeah, it's possible that something like that happened?

huh. right

i merged master into the download_ds branch and now it installed correctly.

so something happened between 1.15 and 1.18.1?

AB is up on bono now but it is returning 502s.

possibly. let's just ignore it :)

yup.

why do we need nginx on bono to run AB?

because we're running it in uwsgi, not the dev server. so we need something that speaks http

ah ok!

i see consul issues in the logs.

it's possible that the last time I ran this was before I made startup changes to AB

probably. so how do we update the consul config here?

the easiest way may be to just have a local config.py in the repo, and then in docker-compose.yml set the `command` to run. this way it won't do any of the prod startup process, and will skip consul

in fact, there is already a config.py, so that might be what I was using before

I've added command: to docker-compose.yml, restarting

http://similarity.acousticbrainz.org:8001/ looks good

well, it's running. doesn't look good :)

thanks!

the logs are clean though.

alastairp: its up here http://bono.metabrainz.org:8001/

it's possible that 1) nginx is trying to serve static files and not pass through to web, or 2) they've not been built in web for some reason

yeah, seems the static files not built. css and js not working.

can we just bring up a dev instance?

yeah, that should work too. just take everything down, find a port that's not used, and consider using the local database so that you have some content

alastairp: the issue with gaga running out of diskspace has to do with the old continuous aggregates. we should really work to merge the time ranges PR ASAFP.

I think the new continuous aggregate will use less diskspace.

[bookbrainz-site] the-good-boy opened pull request #622 (master…Column-min-width): fix(BB-405):Seach: Tabulated results columns have dynamic width https://github.com/bookbrainz/bookbrainz-site/p...

ruaok: ok cool. I'll prioritiese that tomorrow

looks super interesting

alastairp: ok, ping me when you start work tomorrow. lets tag team the PR and release tomorrow.

ruaok, alastairp: any idea about ^?

http is being redirected to https on bono

yes.

that is intended.

in order for letsencrypt to assign certs it needs to have access to port 80 (unencrypted) to start the process. once the process is done, all port 80 requests are sent to 443.

what do you want to do?

but this is running on another port not 80

bring up a dev instance of AB

ah. its a different side of the same setup.

I guess you are connecting to the musicbrainz_docker (or whatever) network? if so, any container that is found on that network will be shunted through SSL.

yeah right. oh ok.

is there a way to get around this?

the setup is quite hacky, it could do with moving to a separate account.

I'll set it up for you/

thanks!

you want to expose port 888, which is HTTP, yes?

yes

ok, you can do it yourself, actually. :)

if you set these 3 env vars on your -web containers, magic will work.

we'll need to setup dns for this. what should we call it.

ab-dev.lb.org ?

don't request that address, yet. don't fill DNS caches with bad data.

maybe dev.lb.org?

i'll probably reuse that for different services then

ok

ok, DNS records set.

go ahead and set those env vars.

then watch the `le` container

on it

nooice!

well, it looked to come up ok.

but page doesn't load.

getting too many redirects error. yeah.

i put this on the web container directly and that'll be running a flask dev server. not uswgi or nginx could that be an issue?

in theory if it speaks HTTP is should work.

i think alastairp mentioned above that nginx is needed for that.

this is what we're using.

ok makes sense. alastairp, had such a setup on bono. it didn't work as expected so i wanted to try a usual dev instance.

i'll try again with his setup.

this is that setup, methinks.

I doubt we have two of them.

do you need access to the MB database for this work

?

if not, simply connected to a different docker network and no magic will happen.

i am not sure, i haven't worked that much with AB before.

let me try without MB db.

ruaok, i moved it to a different network but still it is trying to redirect to https.

if you see the le container still spitting out messages, then you've not escaped it.

2021/05/06 16:46:27 Received event start for container 772bc81ffca8

yeah :(, this is my container

but i rechecked its on a different network

yvanzo, Is it going to be fixed? Do I need to open a ticket to track it?

nelgin, you should probably open a ticket.

_lucifer: I'm hitting my head on an issue with a test, and can't for the life of me understand why it's failing.

If you've got some time, could you please have a look? It's on PR 1390, the error message here: https://github.com/metabrainz/listenbrainz-serv...

sure

Thank you :) I'm slowly going mad.

[bookbrainz-site] dependabot[bot] opened pull request #623 (master…dependabot/npm_and_yarn/underscore-1.13.1): chore(deps): bump underscore from 1.10.2 to 1.13.1 https://github.com/bookbrainz/bookbrainz-site/p...

[listenbrainz-server] dependabot-preview[bot] opened pull request #1431 (master…dependabot/npm_and_yarn/ua-parser-js-0.7.28): [Security] Bump ua-parser-js from 0.7.21 to 0.7.28 https://github.com/metabrainz/listenbrainz-serv...

Project bookbrainz-site build #3953: passed in 4 min 47 sec: https://travis-ci.org/bookbrainz/bookbrainz-sit...

_lucifer: actually I finally figured out the issue, and fixing it now.

Mr_Monkey: oh nice! what was it?