#metabrainz

Hi yvanzo, atj!

I was able to create the db yesterday. However, when I am building the docker compose, I see this error

`failed to solve: rpc error: code = Unknown desc = failed to solve with frontend dockerfile.v0: failed to create LLB definition: rpc error: code = Unknown desc = error getting credentials - err: exit status 1, out: `

I think I am missing some steps. Can you please guide me a bit?

Also, while trying to get it up, I see `Error response from daemon: Ports are not available: exposing port TCP 0.0.0.0:5000 -> 0.0.0.0:0: listen tcp 0.0.0.0:5000: bind: address already in use` How can I change the port?

akshaaatt: https://github.com/metabrainz/musicbrainz-docke...

Thanks lucifer! Worked well for me. 5000 is hardcoded at a lot of places so I had to Replace it everywhere with a different port to get things working.

I think I should have only updated it for the Dockerfile in dev area.

Pretty irritating how Docker registry and MB both default to port 5000

I still don't have the right config yet atj. Please tell me where to update the port exactly.

mooin!

[musicbrainz-server] 14reosarevok merged pull request #2771 (03master…MBS-12740): MBS-12740: Convert Delete attribute page to React https://github.com/metabrainz/musicbrainz-serve...

akshaaatt: `export MUSICBRAINZ_WEB_SERVER_PORT=5555` then run docker-compose

you're running Docker Desktop, correct?

Yes, I have Docker Desktop running, atj

Might be a dumb question, but where do I paste the port updation?

in the terminal window where you invoke docker-compose

did that. Didn't help

Same issue

same `bind: address already in use` error?

Works fine now. I had to restart.

restart what?

The containers

They were already running apparently via another terminal session

when you run `docker-compose` and it fails, the containers started before the error occurred will continue to run

you need to run `docker-compose stop` to stop them

Right. I understand that now.

Thanks a lot atj!

it's a bit confusing and annoying

Phew. It's like doing an intense cardio session just to be able to lift a 2kg dumbbell

I legit want to just update 2-3 css files in MB lol

Finally! I made it! Can work peacefully now (hopefully)

bitmap: MBS-12987 is not really intentional, is it?

MBS-12987: relationship editor pre-fills instrument and vocal roles when adding group member relationships https://tickets.metabrainz.org/browse/MBS-12987

(sorry if I have asked this like four times in the past, was looking through emails)

yvanzo: are you ok with https://github.com/metabrainz/musicbrainz-serve... now? :) (it's now following aerozol's suggestion)

yvanzo: did you receive that email I forwarded (originally bounced) from the Concerty team?

zas: please check support@

forwarded the email for you atj

mayhem: not really an issue of any major importance IMV

that said I think we have correct CAA for other domains, let me check

ok, good to know. but if it is an easy fix, lets do it. so nice to have people contact us on this stuff.

hello,

just wanted to let you know there might be an issus with https://datasets.listenbrainz.org/area-random-r..., it's throwing 504 error

mayhem, atj: fixed, the record was missing for listenbrainz.org, added & tested

nice one zas

if that's the worst they've found mayhem then I'd be very happy!

Indeed!

!m zas and atj

You're doing good work, zas and atj!

I'll respond to the mail.

Already done, I see. :)

about floyd & pink upgrade options, the natural upgrade would be PX93 + hardware RAID10, but it would double the monthly cost compared to current cost. AX102 with software RAID1 would keep cost at ~same level. Since CPU is much faster, RAM is DDR5, and SSDs will be NVMe I don't think performance will be too much impacted (but only experience can tell). There's a few days delay to have one.

I think to go with option 2 upgrade (different IP, run in parallel) (https://docs.hetzner.com/robot/dedicated-server...), I'll coordinate with bitmap for the move (as it will require to switch to new IP).

this way we run benchmarks having both machines in parallel. It will also let some time to ensure the machine is stable. If we detect anything annoying, we'll decide what to do.

bitmap: ^^

Software raid is preferable these days IMO

atj: why is that?

I'm happy that it is, but why?

and if we dont have a lot of disk io (as planned) the software raid should be ok, no?

Hardware RAID is a black box, and can fail in weird ways. Different controllers behave very differently and often don't pass through SMART data on the individual disks so you can't monitor them effectively

ah, ok

Unless you have a memory backed write cache they don't offer much benefit

atj: > if that's the worst they've found mayhem then I'd be very happy!

well, we just received another one. :)

forwarded.

again not really that significant IMV, if we don't have any legacy HTTP endpoints then we can add it

[listenbrainz-server] 14amCap1712 merged pull request #2457 (03master…update-msid-mbid-model): Prevent redundant requests to CAA from taste page https://github.com/metabrainz/listenbrainz-serv...

monkey: LB#2458 waiting for your review.

Use mapping data in recommendation events: https://github.com/metabrainz/listenbrainz-serv...

👍 will look

[listenbrainz-server] 14amCap1712 merged pull request #2458 (03master…update-rec-metadata): Use mapping data in recommendation events https://github.com/metabrainz/listenbrainz-serv...

mayhem: yes, replied to concerty yesterday after releasing fixed mb-docker

Thanks, sorry for not seeing that.

That's because I did neither tell you nor put support in copy.

[musicbrainz-server] 14reosarevok merged pull request #2810 (03master…MBS-12851): MBS-12851: Hide ratings/tags stats for deleted users https://github.com/metabrainz/musicbrainz-serve...

[musicbrainz-server] 14reosarevok merged pull request #2917 (03master…MBS-12749): MBS-12749: Convert cdtoc/attach_filter_artist to React https://github.com/metabrainz/musicbrainz-serve...

HSTS isn't enabled for most websites (it should once we are sure we don't serve over HTTP anymore), atm it is enabled only for picard website, it can be done through docker-server-configs, see https://github.com/metabrainz/docker-server-con...

enabling HSTS for a service still using HTTP is a serious mess...

so I think it can be done for listenbrainz.org (but definitively not for musicbrainz.org)

the report we got by email isn't clear about the domain tested

if we go for enabling HSTS for all websites, we might need to change things a bit to have a short max-age in case of (and later increase it once everything is sure). HSTS is like permanent redirections, it is hard to undo as everything happens on browser side.

I don't think it's reasonable to enable HSTS on all sites, but maybe we can look at what is HTTPS only

zas: ready to help whenever the machine is ready

the plan is to clone floyd first, i.e. have two standbys running side-by-side and compare their performance?

reosarevok: I think the code should probably be taking an intersection of the attributes, rather than a union

bitmap: yeah, that sounds better to me too

bitmap: yes

akshaaatt: lucifer: when you get the time, give the comments a read: https://tickets.metabrainz.org/browse/MOBILE-127

MOBILE-127: Improve listening service to submit "now playing"

jasje: if the user is listening at two places at a time then its fine if either of listens shows as the now playing one

but it would look if we had two cards. It would also help fill space nicely.

look good*

Agh.

I was gonna say the same jasje :p

The idea of a separate now_playing seems like a lot of complication for what seems like an edge case

welp

What do you mean by fill the space? On mobile of website?

lets stick to it then

Wasn’t the now playing thing specific to spotify? I didn’t know other scrobbles provided that feature

website and mobile both

akshaaatt: No, it's just like a regular listen but with `playing_now:true` at the top level

Ah makes sense

let me demonstrate

Then we can have multiple playing nows

jasje: For the website part, don't worry about that, we have some more redesign coming up for the user's listens page

akshaaatt: no, there's only a single playing now at a time. If you send another playing now listen, it overwrites the previous one

listens page mockup https://usercontent.irccloud-cdn.com/file/1Wh2n...

Rough mockup *

yupp I have that figma

👍

I was just suggesting this because there may be mismatch between now playing and bp

We plan on showing listening now and pinned recordings at the top

Between the website and the app you mean?

no the website only

in case lets say someone else (my brother) is listening songs on my phone

that submits a different now playing

but I, on website, sees that my now playing and BrainzPlayer entry is glitching out

(but is actually not)

Well, it would accurately show that you(-ish) are listening to a song on your phone

it does that?

I don’t think we should work on solving this problem. Even netflix accounts are shared by multiple people but all comes under the same person

It can if you submit that information in the listen

In any case that seems like an edge case to me.

alright that solves it

(not in my case, thats a regular case :P)

in my**

I would be fine with keeping an array of playing nows if that is easy enough. Otherwise, I agree with monkey that it is an edge case and we can ignore it

jasje: time to get your brother a listenbrainz account :p

XD

xD

but hell be still using my phone :(

anyways