#metabrainz

I would prefer us to move to HTTP and be done with it.

yes, exactly.

i'm sure zas can provide some reasons why it's not so straightforward

oh, there is server work to be done?

erm, let me check

if it is just the musicbrainz-docker side of things it shouldn't be more than just changing the protocol we use.

yes you're right, http already works

OK, that's the simplest fix then

zas, yvanzo, reosarevok: what's preventing us switching the Docker scripts to pull dumps via HTTP and announcing FTP access to ftp. is deprecated and will be removed in ~3 months?

it would be great if we could get a PR for this fix up today and then merge it in 3 months if we need to the time.

atj: If by Docker scripts you mean musicbrainz-docker, I can look into it.

yvanzo: yes.

We need at least to let the schema change go.

yvanzo: let me forward you the email from atj

meaning we should do it after?

yes, 3 months if 2 months after the schema change, so it's fine.

ah, got it.

this should be a pretty simple fix, right?

Our Docker Compose project supports FTP only atm, but it should be simple to handle HTTP/S too.

mayhem: ^ https://www.irccloud.com/pastebin/JWsQS2dq/

mbdata is using HTTP already

Unless some of our backends need FTP (in which case zas or bitmap would know), I don’t think there is any reason not to drop it.

atj: support for HTTP was a quick hack over FTP (which is here for historical reasons). I see no problem in deprecating FTP, though serving via HTTPS might need some improvements (fancier index pages, ensuring everything is properly configured to serve big files, and better/simpler cert management, etc...)

Well if the status quo is FTP I see no issue with HTTP. HTTPS could be added later.

atj: Is the PASV issue a FTP server configuration issue or client missing flags?

(or both)

yvanzo: it's a client and a server issue but not something that can be easily resolved

The client host is using IPv4 in IPv6 addressing which is something to do with the clients host network configuration

Ok, that’s really easier to move to HTTP/S then.

you end up in a bizarre situation where wget is making a IPv4 request but think it is making an IPv6 request

*thinks

and because FTP is so archaic, they had to add additional commands to the FTP protocol to support IPv6

atj: Are you working on a patch to musicbrainz-docker already?

yvanzo: yes, give me 20 mins

Ok, just let me know if you need any help with it.

reosarevok, bitmap: What is an "MBID submitter"? It’s not documented in https://musicbrainz.org/doc/Editor and there are none in https://musicbrainz.org/privileged

[musicbrainz-docker] 14atj opened pull request #242 (03master…mirror-http): Fetch DB dumps via HTTP instead of FTP https://github.com/metabrainz/musicbrainz-docke...

yvanzo: see above, it's draft at the moment because i'm not sure about the changes to fetch-dump.sh

atj: there are FTP mirrors indeed.

hmm, so what behaviour makes sense? if the URL doesn't start with http:// or ftp:// add ftp:// ?

that should be backwards compatible

lucifer: looks good to me!

who is kleinaviad@.... ?

atj: to add a --base-http-url option and deprecate the current option?

yvanzo: you had to choose the more complex option didn't you ;)

that makes sense, i'll update the PR

mayhem: I doubt that's anyone we know, likely phishing attempt

atj: I think we can merge https://github.com/metabrainz/metabrainz-ansibl... if you have nothing else to add to it.

zas: I want to deploy to sentry first the merge

*then

ok

going to have lunch then i'll do it, are you around?

bitmap, reosarevok: updating MB beta servers

Are you releasing prod after this?

Mbid submitter is unused iirc

It's for some external data users to be able to create their own mbids for stuff they don't want to make public yet

mayhem might remember the details

reosarevok: yes for prod

Thanks

reosarevok: Can you please create the blog post? The editor is not loading correctly for me atm.

updating MB main servers now

reosarevok: pretty much that. still no takers.

reosarevok: on it, seems to work fine with chrome atm

Ok. I can review in a couple hours

(going home now)

zas: bootstrapping sentry - do we want the usual user accounts on it or a subset?

I guess we can have usual ones

ok

🤞🏼

ok, bootstrap done, deploying site playbook

zas: sentry runs the bundled version of docker. the one from the docker repo is newer and has a different package name

not 100% sure how that would go so I think we need to organise some downtime for sentry to do the docker install and a reboot

atj, zas: I updated GH RSA key on clash

bitmap, reosarevok: The script skipped updating MB API container on clash, but there is no expected change in API anyway.

Ok

Same on patton

atj: Is FTP download still mostly working or should we have this patch in today’s release?

yvanzo: it's still working for most people

Ok, thanks. We can also release later today if we want to include this patch in the same release.

I can no more log into paco: kex_exchange_identification: Connection closed by remote host

It's back now.

bitmap, reosarevok: Done with updating web containers; I didn't update others as I manually deployed the patch for bootleg report last week.

atj: I guess we can't avoid a small downtime on sentry

it should be very small

monkey: https://test.listenbrainz.org/user/mr_monkey/ch...

I love how people think that getting a reply within one day is "quick"

shows how poor customer support is in general

atj: amen to that. but we did have a few exchanges before where our side was fast to respond.

I saw your previous response was "lightning" fast :)

atj: I'm dumping all the tax info to our accountants, and I had to admit to our accountants that I failed to pay you last year. And a boozer at my house is not easy to account for, lol. ;)

heh, yeah. I guess everything is different when the people are engaged and proud of their worl.

work,

i encounter the same type of response a lot at work, especially when they ring and a real person without a script answers the phone

yvanzo: i've updated the PR

added a "--base-http-url" argument with a default value, the "--base-ftp-url" argument no longer has a default but if specified has precedence over HTTP

OK, looking into it

zas: forwarded you a mail about picard

hopefully it matches your thinking

taxes submitted. (I love having accountants do the actual work)

atj: It looks neat. Did you test it somehow?

yvanzo: yes, let me add a comment to the PR

hah, I just did another test and it broke :)

forgot to update createdb.sh

made a few comments

Looks good for my stats lucifer

Nice work :)

The real proof will be with user ulkigpulkig and their all_time stats with the two beatles white albums

yvanzo: thanks, maybe we should just add a new --mirror-url option, which can support whatever wget allows, with no heuristics for automatically adding a protocol, and let base-ftp-url override it for backwards compatibility?

[listenbrainz-server] 14MonkeyDo merged pull request #2451 (03master…fix-lb-1189): LB-1189: YoutubePlayer: Use getVideoIDFromListen utility https://github.com/metabrainz/listenbrainz-serv...

that way we don't tie ourselves into HTTP, and can switch the default to HTTPS later without having to add /another/ argument

atj: If adding HTTP option is simpler for now, I’m fine with adding it, releasing it today, and replacing it later on with something more robust.

i think adding a general option is actually easier

I just added a ticket to track the improvement.

👍

[listenbrainz-server] 14MonkeyDo merged pull request #2436 (03master…faraz16iqbal_feature_save-data-header): LB-1011: Don't load album covers if Save-Data HTTP header is on https://github.com/metabrainz/listenbrainz-serv...

atj (apparently this message didn’t go through): We can do that too. These heuristics have been added because many people used to copy/paste URL without scheme.

But there are better ways to handle such oversight.

yvanzo: is the blog ready to review? :)

reosarevok: yes :)

Ok!

(also, fwiw, at least with opera for some reason if you edit a URL to remove or change something, then copy it, it will copy it without scheme)

(no idea why)

We might have a patch for musicbrainz-docker, so release is pending that too.

(release blog post)

Ok :)

But that won't affect the post text, right? Just the link and whatnot

right

We usually remove "Other" from "other task" if there's no React tasks, right?

Made a couple small changes, but it seems good, thanks

Is it fine to merge approved stuff to master already then? :)

Or do you want to finish the docker + post process?

reosarevok: I updated master with production, it is fine to merge stuff :)

Thanks

mayhem: I can take today :)

[musicbrainz-server] 14reosarevok merged pull request #2754 (03master…MBS-12712): MBS-12712: Improve wording for new release adding on disc ID lookup https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] 14reosarevok merged pull request #2761 (03master…MBS-12738): MBS-12738 / MBS-12739: Don't assume "single" and "studio" are ETI unless inside parens https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] 14reosarevok merged pull request #2782 (03master…MBS-12785): MBS-12785: Wrap things sensibly on release sidebar https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] 14reosarevok merged pull request #2898 (03master…MBS-12989): MBS-12989 / MBS-12998: Edit list quick links improvements https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] 14reosarevok merged pull request #2773 (03master…MBS-12779): MBS-12779: Treat "n-bit" and "loop" as ETI if in parens https://github.com/metabrainz/musicbrainz-serve...

ok

:D