#metabrainz

yvanzo, yyoung: https://tickets.metabrainz.org/browse/MBS-11688 seems like something you could look into

MBS-11688: The external link section will disappear once any link starts with //

indeed

ROpdebee: https://tickets.metabrainz.org/browse/CAA-133 I'm sure you know what's going on better than me :D

CAA-133: Several album art upload attempts failed for specific album

reosarevok: Roger that

zas: about https://github.com/metabrainz/picard/pull/1829 , do you mean also expecting uppercase PERFORMER: ? would this happen / be expected?

outsidecontext: nope, but it will be simpler to just lowercase the whole name

ok, yes. got it

whoever invented uppercase/lowercase should burn in hell

reosarevok: that's weird

alastairp: Hi! I forgot to say yesterday that I updated beta.LB. Almost everything is working. There's still one last couple of changes required, as the recommendations page is currently broken.

However, that does allow us to check out what happens if there are no global props. Now we see an error message toast pop up

ruaok: about williams replacement, what about AX51 with 2x8TB (the model just under is 2x2TB)?

monkey: cool, thanks

just going to get a haircut, let's follow this up later

ruaok: same RAM as now, but cpu is much faster (https://www.cpubenchmark.net/compare/Intel-Xeon...)

and much more cores

disks are HDDs with a lot of cache (always on such capacities)

williams is not CPU constrained from what I see, yes?

nope, but that's not a downgrade cpu-wise

right. I'm just trying to get a picture of what williams is doing now.

has spare ram too, so should be fine.

yes, basically we need more diskspace, same or better cpu/ram/network

the price is on par too

ok, sounds good for williams replacement.

which server should be replaced?

we were thinking about boingo, but let me check

for me that was a random example.

it is one of the older server

any of the early servers (ludwig -> boingo) come to mind.

exactly.

PX61-SSD with hardware raid

it was paired with serge

those are more about ram/cpu than disks

prince/boingo you mean?

since they are SSD and not NVMe, those are good canidates, methinks.

prince is quite ilde.

boingo is not busy either.

I think we should move prince to boingo.

candidates: prince, boingo, zappa, trille

I vote prince, because SSD.

ok

everthing else is the same.

On a side note, I jsut updated the hosting wiki and moved lemmy and KISS accordingly.

imho we can drop hardware raid

zas: agreed.

so a better cpu, SSD >= 512gb, and 128gb ram would be ideal

AX41-NVME or AX51-NVME I would say.

actually 64gb ram is enough

ok, I was about to say. 128GB is overkill.

AX51 is ECC ram, bigger SSDs

ok, duh. lets do that.

zas: https://gist.github.com/mayhem/071de0b7c1bc67cf...

looks prince and williams got interchanged.

prince replacement is the targetted future williams (storage)

ah ok!

we move prince stuff to another server (boingo can handled it I think)

order AX51 + 2x8TB, setup, move williams stuff to it

then decommission williams, and replace by AX51-NVME, move stuff from boingo to it

and we do boingo after that?

sure.

should I place the orders then?

yup

ok

so we replace 3 servers, upgrade hardware & software

should I order the 3rd one now or wait? I'm inclined to wait a bit.

according to delays I would say let's order all at once, and manage appointments

up to about 24-26 workdays for AX51-NVME in Germany

I hope it will be faster...

lucifer: alastairp: server upgrades ordered. might take quite some time. we'll keep y'all posted.

👍

ruaok: LB-902 was probably due to ip issue. should i resolve?

LB-902: proxy.listenbrainz.org is not accessible https://tickets.metabrainz.org/browse/LB-902

sure.

monkey: LB-906

LB-906: player does not play any track https://tickets.metabrainz.org/browse/LB-906

money arrived before confirmation arrived. awesome. :)

lucifer: Thanks ! ï can reproduce this if I have both spotify and youtube services disconnected

ruaok: lol, sounds awesome. we have UPI here which works similarly, near instant domestic money transfer.

I'll comment on the ticket

thanks!

The user is still going to see that scary message when connecting youtube?

yeah

zas: no rejections this time!

good :)

zas: is the kiss load check working as expected?

nope, no idea why (yet)

wow, wise is sending money faster than I can order the payments. modern times, finally!

iirc this happened with j5 cluster as well which is on Ubuntu 20.04

defaults are damn low...

off by 100

monkey: what bike stuff were we looking for in the office last week? Just the patch kit?

I don't think I was looking for anything…

will pick one up just in case

you sure were looking in the bookshelf though

monkey: got invoice?

Ah, yes, but not for me IIRC

Err, will do in 10, ruaok

ok, happy to not pay you monkey

oh eh. 🤣

I hope you're paying me euros and not monkeys…

I can pay bananas, its says. compromise?

always a fun half hour on the first of the month: shovel out $30k as fast as I can. lol.

not sure how to fix the check load issue, but if we use defaults params it will break a lot of things (-r divides the load value by number of cpus)

I'd rather define our own command keeping historic behavior

!m zas

You're doing good work, zas!

I updated setup script to set our own default check_load command

ruaok: spark incremental dump fix deployed. i added back the functionality to import all missing dumps but not fail if it can't. weil add the errors next.

nice.

ruaok, what should happen if multiple dumps are being imported and some of them fail. send one mail to observability with imported ones and one mail to exceptions with the errors?

forget sending things to observability, we should aim to get rid of it.

always send to exceptions.

I'd say do as much as possible and report if one or more dumps fail to the exception list.

if everything goes right then no mail?

btw who has access to observability, i am not sure if we even check it so i am fine with removing it

correct.

observability is the wrong concept. Tell us when things work, but keep quiet when they fail?

lets remove it.

indeed. +1

alastairp: when you are available, can you please review https://github.com/metabrainz/listenbrainz-serv... again? except the comment regarding dumped columns, i think its ready.

And for alastairp too: regarding #1420: I fixed the recommendation page blueprint and updated beta.LB

Everything is working now (AFAICT)

yyoung: https://github.com/metabrainz/musicbrainz-serve... seems correct but the test seems to break - are you sure you are trying to find the right thing?

reosarevok: No, that was a mistake, the code of my local master branch is somehow mixed with the UI change I did last week :) I've fixed it and waiting for the tests

Ok :)

Also it seems travis bot wasn't moved here? I didn't see a notification when I opened the PR

I think it was but it's down again

Lemme see