#metabrainz

ruaok: whenever you have some time, do check the "Edit notes linking to a download server with MP3" email and make sure you agree with my take, please :)

reosarevok: tha's it's a bad idea?

morning

!

mo’’in’

moin!

[listenbrainz-server] alastair merged pull request #1215 (master…docs-content-type): Document that Content-Type header is required for submit-listens https://github.com/metabrainz/listenbrainz-serv...

[musicbrainz-server] reosarevok opened pull request #1844 (master…MBS-11310): MBS-11310: Clarify when discid would cause subsecond changes https://github.com/metabrainz/musicbrainz-serve...

does anyone know who the readthedocs user https://readthedocs.org/profiles/metabrainz/ is, and if there's a password somewhere

otherwise, I need to get hold of jfw

(admins on https://readthedocs.org/projects/listenbrainz/)

I think iliekcomputers has access to that

jwf even. helps to spell it correctly

jwf: ping?

shivam-kapila: yes, but it's probably easier to get someone else to do it given how busy he is

alastairp: while you're at it, could you also take a look at brainzutils docs?

in what sense?

I think I tried to set them up, but didn't finish

seems to be some issue with the setup. no docs

oh ok!

well, that's because they don't exist yet

[musicbrainz-server] reosarevok opened pull request #1845 (master…MBS-11309): MBS-11309: Block ToneDen smart links https://github.com/metabrainz/musicbrainz-serve...

[listenbrainz-server] alastair merged pull request #1218 (master…dependabot/pip/flask-testing-0.8.1): Bump flask-testing from 0.8.0 to 0.8.1 https://github.com/metabrainz/listenbrainz-serv...

LB-781 is an interesting LB js test failure

LB-781: Failure in UserListeningActivity.test.tsx due to new year https://tickets.metabrainz.org/browse/LB-781

ruaok: how good are you at debugging DNS?

seeing failures in LB integration tests: https://ci.metabrainz.org/job/listenbrainz-inte...

urllib3.exceptions.MaxRetryError: HTTPSConnectionPool(host='labs.api.listenbrainz.org', port=443): Max retries exceeded with url: /recording-mbid-lookup/json?count=1 (Caused by NewConnectionError('<urllib3.connection.VerifiedHTTPSConnection object at 0x7ff9f15d7fd0>: Failed to establish a new connection: [Errno -3] Temporary failure in name resolution'))

I wondered if it was docker networking issues due to the docker-in-docker setup that we have for jenkins, but I tried to simulate it myself and it seemed to work fine

Zas is better at that i...

alastairp: likely missing dns config somewhere

it'd be intesting to dump content of /etc/resolv.conf

and /etc/docker/daemon.json for docker daemon config

I'll have a look

check if you have any resolver set ('dns' key for /etc/docker/daemon.json)

however I did log into williams, docker exec to jenkins, then docker run a container. and inside there, `ping` successfully resolved the domain name

hmmm, perhaps a python thing then

yes, I'll look at that next

zas: perhaps review the lb.org DNS records? did I make a mistake?

labs.api.listenbrainz.org seems to work and it points to 138.201.227.205

so rather a resolver issue (missing config?)

alastairp: https://stackoverflow.com/questions/16230850/ht... seems to indicate this error can be due to unhandled exceptions (so perhaps the message is misleading)

hmmm ignore, doesn't seem related to dns

zas: yes, there are a number of exceptions there, and it's possible that the cause is due to that

however, tests succeed when run locally, they only fail on williams/jenkins

_lucifer: hi, what do you mean by "the follow server is going to be removed soon"?

I wasn't aware of that, maybe we had some discussions in the office that we didn't share to you about our plans

alastairp: ruaok had shared some days ago that the playlist feature is to replace the follow server

right, that's possible :)

zas: I have reproduced the error. it doesn't appear when I use `docker run` from inside jenkins, but it appears when we use docker-compose and our compose config file

so it's very likely that it's due to the docker-compose network

do you have dns: key in docker-compose.yml ?

no

alastairp: try adding one pointing at 8.8.8.8 to see if it changes anything

yes, I just noticed. jenkins has 8.8.8.8 in its resolv.conf, lb tests have 127.0.0.11

in services: https://www.irccloud.com/pastebin/3vYYi814/

trying now

it doesn't seem to be overriding the dns of the container. I'm looking into it more to see if it's because of our docker-in-docker setup

it's using a new network - jenkins is on 172.17.0.blah, but inside the listenbrainz container it's using 192.168.112.6

I'm going to configure it to use the host network instead of making its own

reosarevok: I kinda think we ought to be more proactive and remove that link from that edit.

Ok, in that case we also need to be more clear about that (IIRC we only mention "don't link to such sites" on relationships)

We have to think about what happens when such a link is being given to prove a bootleg release, too :)

let's be more clear, please. can you adjust the docs and follow up with the user?

Not until we know how to do so

Do we want to block *any* links on any edit notes to any download sites / pages?

zas: our hosting credits with Google cloud have run out and we're unlikely to get any more.

And if so, what do we do with cases where that's the only available source of data

Do we make an exception for those?

going forward lets make sure that we use our resources there wisely and see if we can run most of our stuff at hetzner.

reosarevok: seems fine.

We will also need a way to modify edit notes that doesn't require direct DB editing

But I guess that'd then be my job :p

And I guess a COC of who is allowed to do so (admins?) and for what reasons

(since this isn't something we've actively done before)

jeez, lots of work. :(

I just wanted to remove one little link, lol.

I know :D That's why I was going with "well, it's an edit note, it's not really public unless logged in, maybe it doesn't matter" :p

aight, lets leave it be then.

But long term anyway we *will* want a proper way of doing this (also for like "X insults Y" or even "X doxxes Y" situations)

I'd suggest leaving it be for now, but starting work on a good way to do all this

And once we have that, we can be more proactive and also ask the community to report infringing links for us to remove

We will need to do that if we ever want to open edit notes to logged-out viewing anyway (which is not something we will certainly want, but it's also not something we will certainly never want)

Sounds good to you?

it does, thanks.

good forward thinking!

I also recently wrote some "search in edit notes" code, so that would allow us to find more links if we are pointed to one

(it also allowed me to find a fair amount of notes containing "fuck you" :p Although not all are directed against another person)

wow. the UK won't extradite assange. he got really lucky.

yvanzo: assigned some more https cleanup to you, hope you don't mind :)

Mr_Monkey: alastairp Freso zas : invoices please!

Oh, right. New year.

wait, what day is it?

is it bed time yet?

ah, no not Freso, sorry.

zas: I think I'm going to need your help on this. It's possible that there's a firewall issue preventing things from getting out of williams

"preventing things getting out of williams" seems like an euphemism of some sort for keeping stuff under control

I like it

we're using this docker-in-docker pattern, by mounting the docker control socket of williams into the jenkins container. This means that inside jenkins running `docker whatever` will use the main williams container without having to actually run docker twice

[listenbrainz-server] yvanzo opened pull request #1224 (master…docker-ignore): Docker ignore https://github.com/metabrainz/listenbrainz-serv...

however, when we run docker-compose for our tests, it creates a new network (e.g. 192.168.114.0/20). The gateway of this network appears on williams (192.168.114.1), and I can ping it from the container

reosarevok: no problem, will check after lunch.

I was thinking of flattening the network out, and using the jenkins network as an `external`. However this won't work, because we run the same tests multiple times simultaneously. If we shared a network then we'd have a conflict with the same container names in different test executions

hmmm, I think SEPA and even US wire payments all got faster in 2021. neat.

maybe some banks rolled out sepa instant? transferwise -> bbva is always immediate for me

likely. paying freso got faster. lets see if zas did as well.

zas: let me know when the payment arrives, please.

Mine has always been fast, which is neat

Go go Transferwise

(and, I guess, Power Rangers)

not only that, it's the end of the quarter

go go interim tax returns

[musicbrainz-server] reosarevok opened pull request #1846 (master…MBS-11306): MBS-11306: Update CDJapan normalization and add autoselect https://github.com/metabrainz/musicbrainz-serve...

and end of year. no technical work for me until wednesday at the earliest.

Not a believer in Ballmer peak?

Here all the celebrations are gone. I kinda wish we had Reyes Magos now :p

didn't you hear? we turned them away at the port for not having a negative test

yeah, they're "here", and letter delivery is open, but that's about it

Heh. Two of my friends flew in from Scandinavia and back for Christmas (very responsible of them, I know...)

Nobody even checked their tests or anything

alastairp: ha, that would be the first time someone was *actually* checked at a spanish border that is not an airport.

the requirement of a negative PCR test seems to only apply to people flying.

yeah, I thought it sounded a bit far-fetched even as I wrote it

This was flying, FWIW, so they don't even check that properly

Freso: I'd love your feedback on MBS-11312 and especially on the things that should qualify as "we will remove this from an edit note". I'm expecting "personal info" on self-request, and "spam, insults and piracy" on report, but :)

MBS-11312: Allow admins to modify edit notes https://tickets.metabrainz.org/browse/MBS-11312

Mr_Monkey: "cut short by en dog year preparations" in what dog? :)