mayhem: lucifer there's no disk space left on tito, was something done already?
lucifer[m]
zas: yes, i am looking into it.
zas[m]
ok, thanks
mayhem[m]
lucifer: in the roadmap doc I have this note from you: "Full metadata cache rebuild fails sometimes because we query MB standby/master which have replication lag timeouts" and this is confirmed by the failing cron job -- it fails 100% of the time now and I just realized that what I am seeing and this are connected.
can you tell me more about what is going on? what possible solutions do you see?
lucifer[m]
mayhem: i need to read up on it again but basically whenver replication between the standby MB server lags behind the MB master for more than the specified limit, all active queries are cancelled.
recording cache builds fine, but then the DB connection goes away. in a weird way.
lucifer[m]
MB server logs could possibly tell what's the cause of it.
mayhem[m]
MB PG logs?
vardhan joined the channel
lucifer[m]
yes
mayhem[m]
ok, will look. but the thing is it is trying to make a new connection. there isn't an old lingering connection that dies.
or does PG bouncer hand out a connection that is bad?
I wonder what happens on retry.
lucifer[m]
its possible sqlalchemy is trying to refresh a connection.
yeah, i am unsure what's exactly happening.
mayhem[m]
let me add a retry to it and see if that has any effect.
daleharvey[m] uploaded an image: (322KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/matrix.org/ZJUATdviEusZKafqlfaHzMFg/Screenshot%202025-01-21%20at%2010.53.39.png >
daleharvey[m]
Pretty happy with this, it will go through the full set of listings we have for the city, match them against whats already in the musicbrainz database, warn if the matches arent exact and generate a link to create the event in musicbrainz
holycow23[m] uploaded an image: (13KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/matrix.org/oYgdNTdoLbmmWJCwUzUJEfGo/image.png >
holycow23[m]
<monkey[m]> "I think a mockup might be..." <- Sorry for the delay was a little occupied with something but this is what I was talking about, a simple button linking to the `link-listens` page
Other than that I feel we could actually do a filtering on the link-listens page too based on the filters you suggested, so could you assign me to the ticket and let me know what all filters do you think could be put in there?
mayhem[m]
lucifer: good news: the retry on the bad DB connection seems to have worked:
aerozol, wileyfoxyx, Bob Swift: I am looking at the docs stuff. I understand the hope (at least for mayhem) is to get rid of the wiki entirely, so moving as much as makes sense to the docs (and maybe storing the rest in a wiki-like section in Discourse) might be the best
mayhem[m]
the wiki is a far off secondary goal. Improving MB docs is the really key
wileyfoxyx[m]
I'll be happy with whatever way you'll pick ¯\_(ツ)_/¯
mayhem[m]
lucifer: PR updated.
reosarevok[m]
Whether the docs live in musicbrainz-server is not *super* important at first, being in github makes it easier to remember to update them anyway. The main question is, as aerozol suggested, how much we want to be responsible for
<holycow23[m]> "Sorry for the delay was a little..." <- holycow23: Thanks for the mockup! I think a new button might be a bit too much, but perhaps it would make sense in the existing 'Add listens' dropdown. What do you think?
Also pinging aerozol for his opinion on this (c.f. LB-1722 , and making the link listens page more discoverable)
monkey[m]: Yeah we could cause its sort of related to adding a listen (link) to the MetaBrainz database
lucifer[m]
zas: hi can you please restart `tito`?
monkey[m]
<suvid[m]> "thanks a lot for such a detailed..." <- suvid: Yes you would need to change the .sql definition files too (for new DB creation), on top of having the update script so we can apply the same modification on existing databases.
The .sql files are database definitions for postgres while the models in .py files are SQLAlchemy classes (representing the same data objects) for use in the python codebase.
Otherwise you have an idea of what your database objects look like, but no way to automatically enforce it in python.
<holycow23[m]> "Yeah we could cause its sort..." <- That's also my intuition. Not strictly "adding listens" but definitely related. I think it might be sufficient , but feel free to make a mockup liky you did before and ask the user directly in the ticket, since they are the ones who couldn't find it in the first place.
mayhem[m]
lucifer:
* lucifer: still need a tito restart?
lucifer[m]
yes
mayhem[m]
why not just do a "reboot"?
lucifer[m]
sure can do that. was unsure if something else is needed.
mayhem[m]
there shouldn't be
holycow23[m]
<monkey[m]> "That's also my intuition. Not..." <- Sure, and what do you feel about the flters in the link listens page?
So far, it seems fairly good, although I'm sure we'll want to rewrite a fair amount of the docs eventually
Mostly wondering what's the minimum we need to be able to make this available alongside the wiki while we keep working on it :)
monkey[m]
<holycow23[m]> "Sure, and what do you feel about..." <- holycow23: So you should assign the ticket to yourself to show you are working on it. Regarding the filtering idea, I created a separate ticket for it: LB-1727
reosarevok: I checked and it is getting updated automatically since it's hosted on ReadTheDocs platform
holycow23[m]
monkey[m]: Can you give a little brief about what all would be the filtering options on `link-listens` page?
monkey[m]
Did you read the ticket description? If so, do you have specific questions?
Sophist-UK joined the channel
holycow23[m]
Yes, I did read the description, just wanted to know the options finally allowed to the user is it the three you mentioned track, album or artist names, the default is album so when you go to artist name do you want me to show the entire album and same for the track
atj: This confirm what you thought about API v2: It isn’t stable yet, probably not before Solr 10.
holycow23[m]
Also could you look into the other PRs I had opened earlier, the first was the Calendar Icon, made the changes you requested and the other one was with the config docs
* opened earlier whenever possible, the
monkey[m]
holycow23[m]: Well, I was hoping you could answer that one :)
I haven't tried it so it's hard to project what makes most sense from a user's perspective, and what would be most usable.
holycow23[m]
I feel lets just show them individually, else we are still filtering with album as one layer of filtering
* I feel we should just show them individually in both the cases, else we are still filtering with album as one layer of filtering upon which we apply either of the filters
monkey[m]
I'm not sure what you mean by individually, but I guess meaning not grouped by album?
I think some of it might be clearer once you start on implementation. It's not clear to me how to display the results. For example, the ListenCard component does not show the album name, so if you get a search hit on the album name but only show single listens, there is no clear feedback for the user.
If you show the album group with a single track inside it, then the feature to link the whole album might not make a lot of sense.
It might also be worth trying out the link listens page some more to get some idea of the complications
holycow23[m]
monkey[m]: Cool, I will get working on it, lets go step by step, surely we can discuss, see the results and then come to conclusions with the way to show the results
monkey[m]
Random example: if you search for "Damn", the first track in this group is a hit
monkey[m] uploaded an image: (45KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/chatbrainz.org/STKXzsLhWmjULJtJjWBZDjgk/image.png >
