switching back, cert issues are mostly fixed (I still have to sort out minor issues, but now grade A)
ruaok: the 403s thing was related to the cert issue it seems
ruaok
heh, oops.
zas
curl was impacted ;)
not much an issue, because the used bandwidth by those useless requests is very low
I'll have more corners to round, but the core issue is more or less fixed (Gandi didn't updated their generated cert, which still embeds the expired cert)
[listenbrainz-server] ishaanshah opened pull request #884 (master…master): LB-610: The URL in the browser should update with parameters for the history page https://github.com/metabrainz/listenbrainz-serv...
ishaanshah
iliekcomputers: I have fixed LB-610, I will have a look at the other tickets tomorrow.
the last 4 weeks of data are not on the server, nor are new ones being added.
just testing deduping the data
Zastai
yeah I've been knee deep in Jenkins plugin development; not done much Brainz-related for a few weeks
does that mean that if I reset my timestamp, and redo a last.fm import on test.lb.o, the listen count should stay the same?
(I haven't been scrobbling to last.fm for months, so there's no new listens there)
ruaok
in theory, I believe so.
I'll downgrade that to a solid maybe.
Zastai
Was anything changed wrt avoiding rate limiting on the last.fm side? ISTR always have one or two pages of listens not making it in
ruaok
the importer was just recently redone; there is a good chance it may behave better
Zastai
ok. and with deduping (solid maybe) working, redoing it once or twice isn't a problem
ruaok
can't hurt to try.
though, the counts won't be the same, now that I think of it.
the data pre april 2005 from last.fm is so dodgy that I simply do not import anything prior to then.
Zastai
4750-page import started
ruaok
I might go and take a closer look at it or undo just undo that change.
Zastai
is it the idea that this cleanup (including dropping pre-april-2005 listens) will be applied on the production data?
ruaok
yes.
Zastai
(mainly I'd like to have the last.fm listens cleaned up and redone without losing the spotify ones)
ruaok
I think we share that same goal.
Zastai
(a "delete all last.fm listens" button on My Profile might be nice as a means to testing/redoing last.fm imports without touching the spotify ones)
ruaok
we dont have a good way of identifying all last.fm listens. we do going forward, but not for past data
Zastai
ok. maybe a "delete listens older than <enter date>" button then; in my case, I know when I stopped using last.fm, and I enabled spotify some time after that.
ruaok
I'll just run that query by hand for you. :)
(but I see what you are saying)
Zastai
at page 300/4750 - listens seems to have gone up by ~300. so it's definitely not duping all of them. either there used to be a bug where before the import was missing 1 listen on every page, or there are a handful here and there that aren't getting deduped
does the dedupe logic run per import page, or per listen insert?
(400 pages. increase seems to stay at about 1 per page. i stupidly did not screenshot the count before the import started, so can't tell exactly.)
ruaok
the logic that deduped the production data set for test is different than the dedup logic for normal incoming listens.
the deduper from influx actively seeks out the weirdest cases of the last.fm data.
going forward it will accept listens that are unique on track_name, listened_at and user_name
and now I realize that anyone who ever repeats a last.fm import will just make a hash of this again.
rdswift
Looks like a problem with BrainzBot. Not printing the titles of tickets any longer. Something happen with the recent server maintenance?
Testing... MBS-9009
Zastai
!m BrainzBot staying quiet
BrainzBot
You're doing good work, BrainzBot staying quiet!
Zastai
:p
page 1400; ~1100 new listens, so it's dipped below 1/page
v6lur_ has quit
sumedh joined the channel
ruaok: import finished. 100 listens not imported (expected 237,484 - imported 237,384). listen count went from ~242,500 to 242,931 - interestingly it had previously risen to above 243,000 during the import
i can certainly live with ~400 possible dupes out of 240k
rdswift
What includes (?inc=) to a WS call should I use to retrieve a specific release with the track titles and track artists? Everything I've tried returns the recording title but not the track title.
sumedh has quit
Chinmay3199 has quit
KindTwo joined the channel
KindOne has quit
KindTwo is now known as KindOne
sumedh joined the channel
sumedh has quit
Sophist-UK has quit
ruaok
> i can certainly live with ~400 possible dupes out of 240k
hellz to the yes!
thanks for checking it out. :)
v6lur_ joined the channel
killme joined the channel
prabal
Mr_Monkey: search is not working for the main website