#metabrainz

its nuts

I really need to rework setup scripts, they don't work well enough with recent dists

remember how to fix this?

alastairp: yeah, i reverted it intentionally. there's some issue with consul. timescale not up yet it seems

ruaok: ah, need to run `make`

where?

in postgres-health-check dir, I think

is there a Makefile?

sshing in now

zas: make wasn't installed on gaga, I installed it

tandy: what is nuts?

ruaok: health check is built. try restart?

2021-11-15 16:43:57.979 UTC [99] FATAL: role "listenbrainzts" does not exist

related?

lucifer: that was me.

:+1

alastairp: restart what?

timescale container

are services still down?

reosarevok: just cool that this person is runnin their pc all day to do this massive task

listenstore still down.

wil restart

yes timescale not up acc to consul yet

looks much better now.

typesense also starting now.

yup up now

no errors. fine to continue.

all services up

looking good!

!m lucifer

You're doing good work, lucifer!

tandy: ah, ok. Yeah :D

You linked specifically to a post in the thread about end dates for urls so I was confused :)

reosarevok: doesn't help that discourse automatically updates the url as you scroll, so you have to be aware of that

is empty. and the rabbitmq incoming queue is slowly rising.

oh my bad reosarevok

lucifer: maybe we can restart ts-writer?

yes, lets.

done

there were some errors connecting to rmq in the logs

building typesense now, once that is done will start mapping-writer to clear the other queue

alastairp: true :) tandy: no worries!

queue is falling now

heading home now, back in 30mins in case of emergency or in 1h for meeting. bye

MBS-11545: Genres - partner with RateYourMusic

this ticket is v good idea btw! rym / sonemic has really good tags and mb could benefit from the community's work!

ruaok: should i wait sometime or trigger full dumps now?

is spotify writer re-enabled?

yes

I cant see a reason not to trigger the dump

seems to be working fine.

cool triggering dumps

I was just checking.

ah ok

huh, i do not see the earlier full dumps on the ftp site.

oh! it looks like consul restarted cron container while it was still dumping so the backup dump never happened.

it would be nice if could ask consul to run the terminate_lb_cron script instead of directly restarting the container

I guess that's consul-template instead of consul, you need exec mode for the app to handle kill signal, check kill_signal and kill_timeout (https://github.com/hashicorp/consul-template/bl...)

i see makes sense.

although might be the wrong thing to use here because the use case is to not have consul template restart the particular container while dumps are running.

if full dumps are running, then it might even take a day. setting kill_timeout to a day does not sound good.

rmq queues now clear.

lucifer: yeah, I'm not sure how we can tell consul-template to not restart something

I mean, we could definitely tell it to send SIGUSR1 as a restart command instead, but that'd have a problem if it's telling us to restart explicitly because the timescale service has gone away

indeed.

i wonder if MB does something because MB dumps are long running too.

alastairp: ruaok: btw, weekly checkin before meeting for year in music. up for a quick status update now?

recently they removed a lot of consul service identification and just hard-coded values for config to other services

yep. let's do it.

I was at ismir last week so didn't get a lot done inbetween gaga migration. I'm LB for the rest of the week, so am going to knock out the genre stuff this week

i see. yeah that's one way to do it.

I got top discovery playlists happening again. good progress, but the playlists are kinda crap since a lot of my most liked music didn't get matched. working on making the mapping updatable so that we can rerun no_match results.

Hi there. I'm having problem with search engine. I've updated indexes using sudo docker-compose exec indexer python -m sir reindex followed by admin/setup-amqp-triggers clean & admin/setup-amqp-triggers install & sudo docker-compose up -d and still when i try to search recording (it's in DB) i'm receiving empty array with 'created' field 5 months

ago.

from my side, i was able to get the queries ready for most prominent color and new releases in isolation.

later this week I'll see about knocking out one or two more playlists, but I'd say I'm on track.

lucifer: those look good!

earlier today, also got the script to upload release json dumps to spark so now need to wind this stuff together

after that will start on creating table of LB side.

sounds good.

PetrCBRCZ: perhaps yvanzo can help you

cool, i think we are on track for now. hopefully, a lot of things should be ready to ship next week

Do i need to call any command to make new indexes active ? It looks like even when i've recreated indexes, old ones are still active.

lucifer: yeah, seems likely.

monkey, we'll have some stuff ready soon for you to take over :D

lucifer: when we used to use consul template in the cron container we only used it for rendering the config file, so it had no way to restart the containers. is your setup different?

I wonder how coloful it will be. given how colorful spotify's report is

bitmap: oh actually, you're right. we don't exec cron from consul-template

Is "cololful" when colors look kinda funny?

alastairp: bitmap: uh, interesting. so i wonder how consul restarted lb cron. let me take a look again

monkey: groan

freso: may I go early for I'm making dinner again

restart as in the entrpoint died and docker restarted the container or?

entrypoint*

'consul restarted lb cron' -> consul do not restart any container

ah right. sorry i meant consul template.

alastairp: did you manually interrupt the public dump eariler?

lucifer: yes

and consul-template neither, consul-template might restart/reload programs inside a container though, if configured for that

alastairp: oh ok that must be it. the stack traces threw me off. the last one was broken but scrolling up i see KeyBoardInterrupt before that.

zas, ah i see. makes sense.

<BANG>

It’s Monday of the Imprisoned Writer!

I haven’t received any mailed in reviews, so people up for reviews today are: CatQuest, bitmap, ruaok, alastairp, monkey, lucifer, yvanzo, zas, akshat, reosarevok, Freso – anyone else who want to give review, let me know ASAP. :)

CatQuest: Go!

Meow! `😺`

Last week I was ill :/

so I didn't do a lot of things

I feel better now so hopefully this week I will get back in gear!

plan on doing more stuff and instrument things with reo

♎

fin, go reosarevok

!recall oh no

I'm sorry, I don't remember "oh no", are you sure I should know about it?

!recall oh no.

Hi!

:D

I spent most of the week converting the system for edit display to React

(mostly making sure the flow types work)

noice

And actually releasing the beta / prod releases

Other than that, small fixes mostly, and working on images for instruments with CatQuest

Freso: you? :)

o/

Dealt with reports and flags and such, and also dealing with a persona non grata who has taken to targeting their vile upon me personally.

💪

Other than that, being around and about.

fin.

!m Freso

You're doing good work, Freso!

bitmap: Go!

hey

last week I mostly did code review for the release and worked more on the release editor

I ran into some problems trying to combine knockout and React for the artist credit editing so I'm just converting more of it to React...

I also submitted a PR for MBS-12079 which was reported by Zas to hopefully fix some semantic markup errors reported by Google

MBS-12079: JSON-LD for releases with a lot of [silence] is too big https://tickets.metabrainz.org/browse/MBS-12079

Kill knockout with fire, please

I think that's mostly it. go zas!