#metabrainz

[listenbrainz-android] 14dependabot[bot] opened pull request #172 (03main…dependabot/gradle/io.sentry.android.gradle-3.8.0): Bump io.sentry.android.gradle from 3.7.0 to 3.8.0 https://github.com/metabrainz/listenbrainz-andr...

[listenbrainz-android] 14amCap1712 opened pull request #173 (03main…retrieve-token): Use WebView to extract LB Auth token https://github.com/metabrainz/listenbrainz-andr...

akshaaatt: ^ will need polish probably but works in general.

what?! zas doesn't rhyme with "jazz"???

but that way is how I've always mentally pronounced "zas" :(

well..

i probably pronounce jazz wrong cause it also rhymes with "sass"

I think it does rhyme with jazz? Or Pizazz!

or any other mass, plass, kass etch word

etc*

idk

morena!

Just couldn’t think of anything to go with ‘zas the…’, like ‘Catquest the best’

..

it's CatQuest the Endeavouring Cat!

:P

Haha of course, sorry

oh

uh

it'd only be "the as" and thats not nice

:D

or.. zas the mas-ter

🤷

damn....

reosarevok: I just saw something that may be in relation with the docker image id thing you got yesterday

I just upgraded 1 image, used by 2 containers on moroder, and when I ran docker ps just after, here what I got

both grafana-*-renderer containers are using the same image at this point and were rm -f then rebuilt

but for some reason docker ps shows one with a name, the other with an id....

oh wait, I know why

because the name is shared between both images (something:latest)

so when it displays them, it displays the id of the old image and the name for the new one

it makes sense

Morena

🌅:D

Mōrena!

mayhem: monkey: running behind, as usual, but I’ve put my notes re the recs redesign into this doc: https://docs.google.com/document/d/1N7FmZdJNagX...

So that stuff doesn’t get lost across timelines/metabrainz messages

reosarevok: so, if you have 2 containers C1 & C2 running 'image:tag' with id 1, once you retrieved 'image:tag' with id 2 and used it for C1, C2 would have his image listed as id 1 rather than its name, so basically search by name isn't reliable. And that's perhaps what is happened with your script.

I’ll also put the text from the linked doc here, so there’s a record on IRC https://www.irccloud.com/pastebin/aH3HJjIF/

The doc has screenshots though :)

Thanks a lot lucifer for working on this! ⚡️

I see your PR is in draft currently. Are you working on it or should I pick something up for you?

zas no offence meant. you're the mas-ter of all sysadmin things <3

[listenbrainz-android] 14akshaaatt opened pull request #174 (03main…fix-spotify-remote-crash): Fix Spotify remote crash https://github.com/metabrainz/listenbrainz-andr...

[listenbrainz-android] 14akshaaatt merged pull request #174 (03main…fix-spotify-remote-crash): Fix Spotify remote crash https://github.com/metabrainz/listenbrainz-andr...

lucifer the webview works amazing but lags a lot for me

It basically took 5-10secs to load every single screen and then around 10secs to redirect

akshaaatt: yeah, i had marked it as draft so that you could take over and make any changes you want later. feel free.

O’Moin

akshaaatt: it works quite fast for me even in the emulator.

i am not sure why it lags in your case.

all containers were stopped on kiki & herb

zas: so is the point that while we usually had one container per server for a service, now we have several or?

moin yvanzo :) Seems we need to update the update script

reosarevok: technically you can have multiple containers for a service, even on the same node. It is just about them to not conflict (different ports and names).

Sure, it's just we have never had this issue before the update, so I was wondering what changed :)

(that we'd now have a C1 and a C2 to conflict)

reosarevok: containers update script?

Yeah. It broke when trying to update beta because it was expecting to get back the container name but got the id instead

zas suggested a reason for it a bit earlier

aerozol: I found the PR from the tweet, after checking said tweet's date. I'm afraid it was a no-go: https://github.com/metabrainz/listenbrainz-serv...

Thanks for the feedback document !

Just wanted to point out that we do currently have arrows to the right and left like rob mentioned

Bah, I'll leave comments on the doc

reosarevok: Did you update beta?

No, because the script to update it didn't work, as I said :)

Can you please summarize the course of events :)

yvanzo: you broke it

atj: thanks :D

And reosarevok is giving you both pieces

:p

The docker ps --filter "name=$container" --format '{{.Image}}' thing doesn't work anymore or at least not consistently, it seems

In https://github.com/metabrainz/musicbrainz-serve...

the script compares a container name with a container id

But that used to be both container names

zas said that: "if you have 2 containers C1 & C2 running 'image:tag' with id 1, once you retrieved 'image:tag' with id 2 and used it for C1, C2 would have his image listed as id 1 rather than its name, so basically search by name isn't reliable. And that's perhaps what is happened with your script."

We don't have 2 containers running the same image tag on paco.

Yeah, dunno

Some discussion from yesterday: https://chatlogs.metabrainz.org/libera/metabrai...

zas: Hi! looks like the test websites are working at the moment, at least test.listenbrainz.org, test.bookbrainz.org and test.musicbrainz.org

Requests get timed out.

I looked in more details for BB, the container looks fine (tried restarting it too) but requests don't seem to hit it.

Help please :)

aren't working*

There are new beta images atm but the containers are still running the old image, no longer matching the image beta tag, so it’s expected that the image for the container is listed with an id instead.

Thanks!

Something to start with :)

Yeah, the image was generated, just not updated

But that's the same process we've always followed and this used to work

Which makes me assume the change must be in the new setup from yesterday

Which is why I started by asking zas and atj :)

monkey: let me check

Merci

monkey: re: https://github.com/metabrainz/bookbrainz-site/p... - do you have a table of proposed privs somewhere?

(MB uses 0 as the standard and our flagged privs start with 1, so I was a bit surprised to see 1 as the default)

(although I guess it's not wrong in any way either)

monkey: matching containers properly registered services and they appear in openresty config

reosarevok: Here's what was in the proposal:

But I'm keen on aligning with MB, no reason to be off by one

Oh, ok, so just starting from 1 and there's no "default" no privs flag

I mean, we don't have a 0 flag :)

It's just 0 if there's nothing we would call a priv

I guess 1 is the default no special privs

(so, they just use the site normally)

That's how I read 'ACCESS_THE_ENTITY'

Yeah. I don't think that's wrong either, as long as you document it somewhere

reosarevok: I just ran the script again and it worked fine.

= read

yvanzo: huh, ok :) Maybe I should have just retried :D

"retry until it works" is not the most ideal approach, but hey, if it works for our selenium tests 🤣

atj, zas: Might it be that docker-server-configs working copy at paco was not up-to-date when reosarevok tried to update beta?

zas thanks for checking. Mystery. I swear test.LB and test.BB were working yesterday after the consul migration, I checked them

monkey: do you have a constants file? I'd put the constants in that file already if so, so that everyone has a clear view of the meanings from the get go :)

Since the script that failed is the one that zas patched in the PR. The error message was possibly about the consul container being stopped.

monkey: I keep searching, the problem is real, that's just not that

Yep

yvanzo: ok, I guess I'll continue with a prod release then for now and see if it all works, unless you have anything against it? :)

Well, after a brief check that beta works :)

reosarevok: Good call. We don't have a central constants file yet but there's a good place for it

GET https://staticbrainz.org/MB/common-ba8f919.css net::ERR_TIMED_OUT

zas: is there some expected issue with staticbrainz?

I see a few websites went down in the past coupl eof minutes

(CB BB LB)

Could be related

reosarevok: Do you have the full log of this failed attempt to update beta?

yvanzo: sending it in private :)

something weird and bad is happening

checking

Sorry, it was already the full log.

2023/06/07 09:10:43 [error] 267#267: *7374 lua tcp socket connect timed out, when connecting to 10.2.3.1:6379, context: ssl_certificate_by_lua*, client: 10.2.2.60, server: 0.0.0.0:60443