Does InitDb.pl have a setting that allows you to change the postgres schema it initializes? If not, is this something I could contribute?
lucifer[m]
monkey: BookBrainz was down, didn't find anything in `bookbrainz-web-prod` logs but restarted the container and things seem to work fine. would be great to add timestamps to the BB webserver logs.
m.amanullah7: oauth tokens are meant to expire soon, update your code to refresh the token if expired and see if that helps. otherwise its a different issue and i'll take a look.
oh i see, you are getting the error during the refresh token flow. push your changes, i'll check locally.
mamanullah7[m]
lucifer: Sure!
suvid[m]
lucifer: i think i'm pretty much done with everything now
just have to implement submit listens function now
is there any already existing function i should use for that?
mamanullah7[m]
<lucifer[m]> "oh i see, you are getting the..." <- lucifer: i've deleted the refresh_token func while refactoring! i'll add and will try once then will let u know!
yvanzo[m] joined the channel
yvanzo[m]
Hi zas, we can shut down Solr 7 cluster as of now.
zas[m]
yvanzo: ok, I'll proceed
yvanzo[m]
zas: I’m working on stopping the indexer.
(and removing the triggers, and…)
yvanzo[m] is off for lunch
petitminion joined the channel
julian45[m]
zas: once solr 7 instances are shut down, could the servers perhaps be briefly kept around for use as a [nomad](https://developer.hashicorp.com/nomad) testbed?
zas[m]
We get 502s on mb right now
yvanzo[m]
I just tried to removed SIR triggers from PG.
-d
zas[m]
yvanzo: it's back, likely temporary due to triggers removal
julian45: those are VMs, if you want to conduct tests we can easily create new ones for this purpose
julian45[m]
zas[m]: fair
yvanzo[m]
zas: OK, now resuming triggers removal (90% left).
I canceled it again. Got many ERROR: canceling statement due to statement timeout
reosarevok[m]
I guess we need to temporarily increase the timeout time?
yvanzo[m]
Running again.
Same issue again.
It just removed 3 triggers, there are ~300 left.
I stopped it again.
I didn't expect it to cause that such a slowdown to the website and API.
Trying it again with a commit.
OK, same effect, aborted.
Need to announce the downtime as I don’t see any workaround.
zas: I just stopped the indexer for now.
zas[m]
Can we still delete VMs?
yvanzo[m]
Yes
Index messages will just queue up in RabbitMQ for nothing.
petitminion has quit
yvanzo[m] is off now
zas[m]
I stopped them
lucifer[m]
<suvid[m]> "is there any already existing..." <- take a look at api_tools.py or the existing listens importer, they have a function to submit a list of listens using rabbitmq
it looks like there's been a good bit of work on title case caps and I found an edge case that probably should be addressed since it likely comes up a lot in songs: `Les Brers in A Minor, Pt 2` looks like "Minor" should stay caps from what I've seen but the auto correct is changing it to "minor"
it doesn't make sense anyway since Minor's not a preposition.
Hey ansh: i was going through the `brainzplayerSettings.tsx` to add funkwhale in code i saw we are hardcoding priority in default!! thats not the issue but i user already selected their priority list if new externalservice is added by default it will not be added in user priority list until he select again! let me if i misunderstood anything>?
<mamanullah7[m]> "Hey ansh: i was going through..." <- i was thinking of dynamically pushing the newly added in existing user priority list? is it worth it! to add remove from user priority if any services removed or added?
monkey, ansh or anyone who can suggest about this!!
yvanzo[m]
bitmap: Do we have any kind of priorities for statements/users in our PG instance? Could it be used to preserve MB response time while deleting the old triggers and functions used for live indexing the now-deleted Solr 7 cluster?
bitmap[m]
yvanzo: I don't think PG has this functionality. strange that it causes timeouts though, I wonder if that requires an exclusive lock on the table (the PG documentation doesn't mention it)
you could try disabling the triggers for now instead
Hey mayhem: Is it okay if i use [hset](https://github.com/metabrainz/brainzutils-python/blob/bf01c6da15d4a2426d64a31cf232c06bec3860f3/brainzutils/cache.py#L279) to store failed messages in the redis queue.
Im thinking of using failed-notif as hash name and notification ID as key and the notification data as value.
That way, when we need to retrieve all failed messages from the queue, we can do it with hgetall.
s/the//
yvanzo[m]
bitmap: Thanks, I disabled the triggers at least.
Now deleted!
bitmap: I have to update the sys admin docs but it's going to conflict with your current PR.
bitmap[m]
yvanzo: oh no, let me apply your suggestions and merge that
yvanzo: merged now
petitminion has quit
DjSlash has quit
DjSlash joined the channel
yvanzo[m]
I will complete that on tomorrow.
Jigen
bitmap[m]: THANKS your tine bit ofcode is going to save me *so* much work <3
bitmap[m]
hehe no problem
Kladky has quit
aerozol[m] has quit
<nik[m]> "Does InitDb.pl have a setting..." <- well, there are multiple schemas that get initialized: musicbrainz is the main one, but there's also cover_art_archive, event_art_archive, statistics, documentation, etc. customizing any of these isn't supported because they're hardcoded throughout the code and SQL scripts (so even if you don't run the server, the replication scripts and schema upgrade scripts would also fail)
what is the use case you have in mind?
iconoclasthero has left the channel
nik[m]
Oh, I forgot about all the secondary things. For my app, I only use the big musicbrainz one. My goal is to be able to easily swap out a running dataset with a fresh one. I was planning to load a fresh one into something like musicbrainz_new and then transactionally swap it.
* Oh, I forgot about all the secondary things. For my app, I only use the big musicbrainz one. My goal is to be able to easily swap out a running dataset with a fresh one. ie, loading into musicbrainz_new and then transactionally swapping it with musicbrainz.
It was initially setup with `mbslave`, but that project is not maintained and that's starting to be a dealbreaker for me.
bitmap[m]
nik: I see, we could consider accepting a patch but it'll be difficult to replace all of the hardcoded references
nik[m]
I agree, it would be quite difficult, especially considering how many there are
bitmap[m]
if you can manage a way to change the search_path for you app at runtime you could also do it in two stages (copy musicbrainz to musicbrainz_old first)
yeah...you'd have to either make that configurable, or remove the schema name from your queries and configure search_path on the PG role/database level (which you can then change on the fly, you'd just have to make sure all existing sessions are referencing musicbrainz_old before dropping musicbrainz)
rostiku[m] joined the channel
rostiku[m]
careful with merging with the musicbrainz database, I had more luck creating a complete separate one
this way you can just sync with musicbrainz-server whenever you want
