OK, easy fix, I'll get this done for the release that I'll do with iliekcomputers this/next week
yvanzo
I had an issue with using sir/solr images locally and found out it was because of consul-template-base but I might be wrong after all.
alastairp
there's a change between the old and new consul base images in the way that we configure stuff - metabrainz.org uses the new one, I believe AB still uses the old one
i wanted to ask can we start offering the app through metabrainz google play account. you won't need to provide me access to the google account. I can just be given limited access to manage only the MusicBrainz for Android app.
yvanzo
or wait if still needed, yes
ruaok
it doesn't. :(
yvanzo
ruaok: oh uploaded version is missing dash '-' after 'v'
actually, the two latest tags have this issue.
uploaded v2019-05-22.1 whereas it should be v-2019-05-22.1
reosarevok: np, already checking all the small ones I see that don't have changes requested
reosarevok
Thanks!
ruaok
iliekcomputers: this problem is happening for other users as well. :(
reosarevok
Ban all the users!
alastairp
column on profile table that says if they're string-endian or int-endian and cast accordingly?
ruaok
lol
replaying the user would be the best thing.
alastairp
yeah, sure
ruaok
iliekcomputers: your fix does work, but the influx-writer crashes (but skips) the bad users listens.
far from ideal, but we are catching up now.
the osvo-fix has been reverted.
yvanzo
bitmap: Can we retag v-20110711-schema-change to v-2011-07-11-schema-change? It is the only mistagged release so far, always pops up first, and has no link to it from either blog or tickets.
ruaok
I'll update tickets appropriately next.
yvanzo
bitmap: Also it would be nice if we could add patch version .X as of the next release (btw, that would align with other MeB projects).
bitmap
yvanzo: sure
yvanzo
That would also help mirror maintainers with following releases and hotfixes.
yvanzo: doesn't the same issue exist if they switch tags? unless we announce every hotfix and patch version on the blog with upgrade instructions
which might be a good idea, but it's extra work and my thinking is that tags are only useful if people use them :)
(I mean, we should definitely mention in the blog if a release requires some change in dependencies already)
yvanzo
bitmap: Hotpatches usually don’t break anything, but further releases may. MB VM relies on MBS release tags already.
bitmap
in my experience most of our hotfixes are for display/editing issues which are less likely to apply to someone using a mirror server, so it just doesn't seem like every hotfix necessarily needs a tag (though I agree some changes should)
if the MB VM needs a tag for a certain patch then by all means we should add one
I just wanna make sure someone/something will actually use them before changing my workflow for hotfixes
like, a patch version to hotfix an issue in the webservice makes total sense to me
gr0uch0mars has quit
iliekcomputers
ruaok: influx-writer is now skipping the listens for the users with type mismatches
but it doesn't seem like influx-writer is restarting...
ruaok
yep. see tickets above.
but something causes it to slow down quite a lot everytime it encountered a bad user.
restart or delay otherwise, our pipeline is freaking slow.