#metabrainz

akshaaatt: Hi! Just letting you know I've deployed the changes we talked about (apart from Recent page) for the empty column on test.LB for easy testing of the associated PR https://github.com/metabrainz/listenbrainz-serv...

Thanks monkey!

monkey I think the mobile version should hide the followers, etc

Mobile UX is highly affected by the change

[musicbrainz-server] 14reosarevok opened pull request #2442 (03master…MBS-12231): MBS-12231: Don't try to load last annotation if there's no entity https://github.com/metabrainz/musicbrainz-serve...

yvanzo or bitmap: ^ let's make sure we merge this for next beta. zas: thanks!

Makes sense

akshaaatt: I pushed that change but haven't updated test.LB yet

Hi monkey! Approved the PR 💯

Also, are we only supposed to let the author merge the PR or the reviewers could merge it as well?

I think letting their authors merge is probably the safest, considering I might have found a last minute bug or something and didn't mark the PR as not ready

Yay!

Alright, makes sense to me.

[listenbrainz-server] 14MonkeyDo merged pull request #1890 (03brainzplayer-redesign…monkey-empty-bp-column): Fill right hand side column https://github.com/metabrainz/listenbrainz-serv...

monkey I think we can also merge the bp redesign branch to master now?

A prod release would be good as well :)

I have another PR on the way to deal with the Recent page (we wanted to add some sitewide stats), which has a medium-sized refactor rolled into it. But next week we can merge and deploy the new BP without there being a hole left anywhere :)

Oh right! Sounds nice

alastairp: lucifer: yeah the 9.5 -> 12 upgrade was in-place, I had written a script to do it at https://github.com/metabrainz/docker-postgres-c...

thanks bitmap

the google doc we had was https://docs.google.com/document/d/1VXC_A8kj74x...

looking at the steps, we did a barman backup a few hours before, and then once traffic was stopped we did a local backup of the data dir

the barman backup was sufficient since it continues to save any subsequent wal files that are generated, but the local backup was just to allow restoring much faster than over ssh

and the idea is that an in-place upgrade is always going to be faster than a re-import?

has MB ever had a problem with an in-place upgrade? last time I checked, the official explanation from pg was "it works well, but we don't guarantee anything"

hmm, well there was one weird bug related to the cube or earthdistance extensions which we had to workaround, but we did that before the upgrade

the pg_upgrade would fail without that

but yes, the in-place upgrades are very fast

I didn't get that impression (that it may be less stable) from their documentation at least

[bookbrainz-site] 14the-good-boy opened pull request #793 (03master…collectionWorkTableFix): fix(Collection): conditionally render author name column in the Work Table component https://github.com/metabrainz/bookbrainz-site/p...

hmm, right.

maybe it was something like this that I remember seeing: https://www.postgresql.org/docs/9.1/pgupgrade.html "If a future major release ever changes the data storage format in a way that makes the old data format unreadable, pg_upgrade will not be usable for such upgrades."

makes sense

there are also some warnings about your old cluster being unusable once you use --link

apparently --clone doesn't have that issue but is only supported with btrfs and xfs

[listenbrainz-server] 14MonkeyDo opened pull request #1891 (03brainzplayer-redesign…monkey-refactor-recentlistens): Refactor RecentListens component https://github.com/metabrainz/listenbrainz-serv...

link uses a hardlink for the data files in the new directory?

^That's the PR I was talking about akshaaatt, splitting RecentListens into two components, one for user's listens and one for the "Recent" page. sitewide stats in Recent will be in a follow-up PR.

yeah

see also "Reverting to old cluster" on https://www.postgresql.org/docs/12/pgupgrade.html

this says you can still recover the old one (by moving a file) as long as you didn't start the new one yet

cool, so we should check the compatibility of timescale with this, but it seems like it's something we could use for an upgrade

bitmap: the script and doc are useful. thanks!

in floyd's case (or was it bowie back then) I think I used pg_dumpall with --schema-only on the prod cluster to get an exact copy of the prod schema, then I was able to quickly test the upgrade locally

but eventually we made a complete clone of the prod cluster onto williams and tested the upgrade there too

bitmap: early next week, we'll upgrade burnside, there are instances of redis (and few more stuff) running on yehudi I wish we move to the new server (burnside2), in order to retire yehudi. Can you have a quick look and tell what it implies? I think those redis instances are used by mb.

there's also musicbrainz-sitemaps

the sitemaps and redis cache containers should be trivial to move at least

there is also redis-store which contains sessions, those will have to be backed up & restored on the new server, but not too difficult

ok, thanks, so nothing too hard to move to the new server. We'll have some time to proceed, but I want to cut costs related to yehudi as soon as possible. We'll try to keep that slot free to ease server moves in the future.

btw, burnside is an AX50, it will be replaced by AX51 which are much more powerful (more cores mainly)

I increased the weight of aphex for mb web & ws, it handles more queries than any other server, without much load (under 4 atm)

[bookbrainz-site] 14MonkeyDo merged pull request #793 (03master…collectionWorkTableFix): fix(Collection): conditionally render author name column in the Work Table component https://github.com/metabrainz/bookbrainz-site/p...

nice

aphex is AX51 too?

yes

burnside2 will be the exact same hardware

[bookbrainz-site] 14MonkeyDo merged pull request #792 (03master…helpPageDiagram): fix(static): Update entity-relationships diagram on the Help Page https://github.com/metabrainz/bookbrainz-site/p...

[bookbrainz-site] release 03v-2022-03-04.0 has been published by 14release-drafter[bot]: https://github.com/metabrainz/bookbrainz-site/r...

monkey: to confirm RecentListens.tsx in the PR is listenbrainz.org/recent ? and Listens.tsx in the user listens page.

That is correct

👍

The name was always a bit confusing 😅

PR #1891 (RecentListens refactor) deployed to test.LB

Updating BB prod

Hello everyone. Hope you all are doing good.

I was trying to implement a menu that shows pages and lets us select one of them, in the Search Activity. For that purpose, I need the responseObject count, or the size of the data list. This I can get from SearchPagingSource, but I am concerned about how can I use this variable from SearchPagingSource class in SearchActivity.

Given, the variable we are getting is from the API request's response.

akshaaatt: how should I approach this poblem , like what should be the generic method to tackle such problems?

Should I use sharedPreferences in such cases?