hi alastairp: Sure, this afternoon would work for me, however the new instance is not available at the moment.
2022-03-01 06052, 2022
reosarevok
bitmap: when around, let's talk about that ^ for the actual names of things like RT_SLAVE and indices and whatnot, see what can be changed and what not etc
mayhem: for listen count of current month, i am thinking just to use timescale. it'll use 6-7 chunks at most so should be much of an issue. we can cache it for some time as well. alternatively, we can calculate in spark in store as a stat LB side.
2022-03-01 06027, 2022
mayhem
seems sensible
2022-03-01 06039, 2022
lucifer
👍
2022-03-01 06017, 2022
BrainzGit
[bookbrainz-site] 14MonkeyDo merged pull request #787 (03master…dependabot/npm_and_yarn/react-datepicker-4.7.0): chore(deps): bump react-datepicker from 2.16.0 to 4.7.0 https://github.com/metabrainz/bookbrainz-site/pul…
2022-03-01 06045, 2022
alastairp
thanks yvanzo. I had some other things come up this afternoon. I was looking at some CB PRs and saw that there will be some significant translation changes to be made. If you can ping me when the instance is ready then I'd be happy to help to move CB translations to it. I'm in no hurry to do so
2022-03-01 06045, 2022
yvanzo
alastairp: Thanks, will keep you informed of any update. I hope it will be in the next days as a new version of Weblate has just been released yesterday.
2022-03-01 06058, 2022
alastairp
Ansh: regarding the missing data for the sample database, I'm thinking of two options: one is that for a development database (e.g. DEBUG=True, or another setting) we can construct dummy metadata for items that are not in the database. To do that we just need an extra check here: https://github.com/metabrainz/critiquebrainz/blob…
2022-03-01 06032, 2022
alastairp
another option is that we create dummy reviews for all of the entities that are in the sample MB database, that someone can import if needed
2022-03-01 06042, 2022
alastairp
lucifer: your feedback on this would be welcome too ^
2022-03-01 06053, 2022
lucifer
why do we need separate behaviour for dev?
2022-03-01 06030, 2022
alastairp
because now we hide reviews if the reviewed entity isn't in the MB database. and if the developer only has the sample MB database instead of the full one, they can only see about 6 reviews out of the 5000+
2022-03-01 06027, 2022
lucifer
oh so this is for importing review dump and MB sample dump?
2022-03-01 06038, 2022
alastairp
yes
2022-03-01 06035, 2022
lucifer
i guess do nothing or always show a fixed name for those reviews in dev.
2022-03-01 06057, 2022
alastairp
yeah right - my first suggestion is the same as your second option
2022-03-01 06028, 2022
alastairp
`if config.get('DEV_HAS_SAMPLE_DB'): <fill in missing metadata with dummy value>`
2022-03-01 06006, 2022
lucifer
yup makes sense
2022-03-01 06015, 2022
alastairp
cool, thanks
2022-03-01 06053, 2022
reosarevok
Seems sensible to me
2022-03-01 06007, 2022
reosarevok
Do you have a sample DB for CB? I guess there's not that many reviews to need it?
2022-03-01 06040, 2022
alastairp
reosarevok: right, the db dump is <10MB compressed...
alastairp: I had submitted a pr for removing Gravatar. Can you review that?
2022-03-01 06026, 2022
alastairp
Ansh: yes, I already looked at it! I have some comments to make on it, which I'll do when I've finished my other tasks for today
2022-03-01 06000, 2022
Ansh
Okay.
2022-03-01 06019, 2022
alastairp
there is a database field, `user.show_gravatar` where users can choose if they want a gravatar to show. We can also remove this (and the associated config item on the user's profile page)
2022-03-01 06045, 2022
Ansh
yes, I saw that. I think we can leave it as it is. If we want to integrate some other type of avatar in the future, it would be easier
2022-03-01 06014, 2022
alastairp
we removed gravatar completely from musicbrainz, so I think we can also remove it completely from CB
2022-03-01 06008, 2022
Ansh
I saw the PR for MBS-12137. It said that support of avatar is still kept in the codebase if we want to integrate in the future.
we want to use user avatars from the community.metabrainz.org forums, when we work out how to do that we can integrate it, but because it's all controlled completely within our infrastructure we can probably just enable it by default
2022-03-01 06019, 2022
alastairp
ah, interesting :)
2022-03-01 06054, 2022
yvanzo
actually we replaced avatars with generic editor icon, not to change the page layout
2022-03-01 06038, 2022
alastairp
yvanzo: mmm, right. There are some users who do have specific gravatars: https://critiquebrainz.org/ (aerozol and jesus2099)
2022-03-01 06010, 2022
alastairp
but I'm not sure if we should also replace it with the generic MB user avatar, or remove it completely. I think it's OK to remove it completely
2022-03-01 06032, 2022
lucifer
+1 to remove completely
2022-03-01 06046, 2022
alastairp
I can't remember the steps that are necessary for adding discourse avatars
2022-03-01 06002, 2022
lucifer
if we need this code in future, its in git history.
2022-03-01 06012, 2022
alastairp
+1 absolutely
2022-03-01 06028, 2022
alastairp
I think that a single editor icon (like the MB one) would make the front page or browse page of CB quite repetitive
Ansh: by the way, the work that you have already done on that PR is great. thanks!
2022-03-01 06042, 2022
dseomn joined the channel
2022-03-01 06045, 2022
bitmap
zas: ping. thanks for checking. I don't think we're running the latest PG point release either, so we should upgrade that too just in case
2022-03-01 06010, 2022
zas
bitmap: can we move active db to pink in order to proceed? When this would be possible?
2022-03-01 06007, 2022
bitmap
yes, we could try this, but it'll require downtime so perhaps when there's less traffic
2022-03-01 06039, 2022
bitmap
the repmgr failover itself isn't instant, and also the service names need to be swapped so that pink is {postgres,pgbouncer}-master and floyd is -slave
2022-03-01 06003, 2022
bitmap
do you want to attempt this today or tomorrow?
2022-03-01 06034, 2022
zas
tomorrow
2022-03-01 06015, 2022
zas
we need to plan it carefully, I think we should upgrade the system as well while at it.
2022-03-01 06058, 2022
zas
but only if hardware checks report no issue
2022-03-01 06011, 2022
bitmap
okay, I can upgrade the postgres cluster repo and prepare some steps for the failover
2022-03-01 06032, 2022
bitmap
what would be a good time to start? I can make a blog post too
2022-03-01 06033, 2022
zas
at which time will you be available tomorrow?
2022-03-01 06055, 2022
bitmap
normally from around 16 UTC but I can appear earlier if needed
2022-03-01 06000, 2022
bitmap
reosarevok: wanna chat about the replication terminology?
2022-03-01 06006, 2022
bitmap
zas: somewhere between 14 and 24 UTC would be ideal for me I guess
2022-03-01 06041, 2022
yvanzo
bitmap: Do you have some time to discuss dbmirror2's implications w.r.t. sir?
2022-03-01 06052, 2022
bitmap
yvanzo: sure
2022-03-01 06047, 2022
bitmap
my initial idea there was that we'd mirror the dbmirror2.pending_data data for sir's usage
2022-03-01 06020, 2022
yvanzo
bitmap: do you mean that sir would query a pending_data table directly?
2022-03-01 06035, 2022
bitmap
yes, since it would contain all the information it needs, so there wouldn't be any need for triggers to push the changes into a queue
2022-03-01 06003, 2022
bitmap
the pending_data mirror would act as the queue
2022-03-01 06034, 2022
yvanzo
bitmap: There are theoretically some benefits with using rmq but I guess it's not really used as it should (no message deduplication atm for example).
2022-03-01 06055, 2022
yvanzo
bitmap: Querying a table instead of a queue is already working for mglubb, so there is some code available to make it work for sir. :)
2022-03-01 06008, 2022
bitmap
right, I don't mean rmq isn't useful, just not really in this context, there's not enough traffic or need to have complex routing
2022-03-01 06008, 2022
bitmap
oh yes that's convenient to have something pre written then :)
2022-03-01 06035, 2022
bitmap
(we can also deduplicate things on the sir side in the future - just pull in a chunk of changes at a time)
2022-03-01 06055, 2022
reosarevok
bitmap: sure, I can
2022-03-01 06013, 2022
reosarevok
(sorry, was watching news)
2022-03-01 06014, 2022
bitmap
hey
2022-03-01 06020, 2022
bitmap
np
2022-03-01 06025, 2022
reosarevok
So yeah, if the idea is to move from slave to mirror
2022-03-01 06047, 2022
reosarevok
We have a lot of triggers, RT_SLAVE on dbdefs, etc
2022-03-01 06005, 2022
reosarevok
Which of those require a schema change to rename, and should we do it during this schema change?
2022-03-01 06037, 2022
bitmap
for RT_MASTER and RT_SLAVE specifically I think we can introduce new constants like RT_PRIMARY and RT_MIRROR. but I'd keep the old ones around for compatibility -- at least for a while
2022-03-01 06007, 2022
bitmap
this shouldn't cause any breakage then but we can encourage people to change them in their settings
2022-03-01 06016, 2022
bitmap
in the upgrade notes
2022-03-01 06013, 2022
reosarevok
I wouldn't change MASTER yet, personally, but I guess we can if we must
2022-03-01 06036, 2022
bitmap
sure
2022-03-01 06049, 2022
reosarevok
Anyway, the basic principle would be the same
2022-03-01 06023, 2022
reosarevok
In the code we have a lot of references to slave but that is probably not a hard change
2022-03-01 06046, 2022
reosarevok
But that's why I was asking about the schema change stuff, which is harder
2022-03-01 06054, 2022
bitmap
well, renaming the PG functions/triggers with slave in the name can be done this schema change if we want, and shouldn't break anything
2022-03-01 06010, 2022
yvanzo
👍
2022-03-01 06027, 2022
reosarevok
Would that just involve "drop them with the old name, create them with the new"?
2022-03-01 06055, 2022
bitmap
yes
2022-03-01 06009, 2022
bitmap
iirc there is a way to rename a function, but
2022-03-01 06024, 2022
bitmap
any functions that call it also have to be redefined
