Oh, good that we already have some new relationship types.
2014-12-01 33517, 2014
ianmcorvidae
since the first migration failed but postgresql sequences aren't transactional and we weren't setting the sequences before inserting into the link_type table for the master migration, there's a big gap in the ID numbers on production
2014-12-01 33530, 2014
ianmcorvidae
and of course all the rows of the link table reference the production ones
2014-12-01 33531, 2014
Freso
Oh. Meeting time, eh?
2014-12-01 33525, 2014
ianmcorvidae
however, at least the link_type bit we know what's affected clearly, the CAA stuff needs some comparison of databases
2014-12-01 33511, 2014
ruaok
ugh. how can we go about fixing that?
2014-12-01 33511, 2014
ianmcorvidae
just deciding exactly what way to fix it (moving slaves to match production or the other way around)
2014-12-01 33526, 2014
ianmcorvidae
well, the same way we've been fixing this sort of thing elsewhere
2014-12-01 33550, 2014
ianmcorvidae
one way or the other it should be possible to create a replication packet that removes or updates the relevant pieces
2014-12-01 33521, 2014
ruaok
oh, thats pretty clever.
2014-12-01 33530, 2014
ruaok
pain in the ass, but clever.
2014-12-01 33549, 2014
ianmcorvidae
moving slaves to match production would be: disable replication triggers and FKs to the table, insert fake rows in all the mistaken gaps; turn back on replication triggers, delete all the rows (including the correct ones), then reinsert the correct rows only
2014-12-01 33553, 2014
ianmcorvidae
and then turn back on FKs
2014-12-01 33520, 2014
ianmcorvidae
we did this for wikidocs/stats/documentation (with less complication since those don't have FKs pointing to them, nor needed inserting fake rows) the other week
2014-12-01 33547, 2014
ianmcorvidae
CAA will require figuring out how to not screw up the index.json generation process and not accidentally delete any images or such :P
2014-12-01 33524, 2014
ruaok
yikes.
2014-12-01 33538, 2014
ruaok
is this the main thing you're currently focusing on?
2014-12-01 33542, 2014
ianmcorvidae
after we figure out exactly which rows of CAA stuff need fixing
2014-12-01 33557, 2014
ianmcorvidae
other than doing a release today, probably
2014-12-01 33528, 2014
ruaok
ok, at some point very soon we need to go cover the JSON-LD proposed changes from google.
2014-12-01 33542, 2014
ruaok
they sound reasonable, but we need to go over stuff.
2014-12-01 33503, 2014
ruaok
can we shoot for wednesday?
2014-12-01 33510, 2014
ianmcorvidae
sure
2014-12-01 33532, 2014
ruaok
k. I'll keep out of your hair for now. :)
2014-12-01 33544, 2014
ruaok
ok, anyone else got review bits?
2014-12-01 33534, 2014
ruaok
right then. onward.
2014-12-01 33540, 2014
ruaok
alastairp: present?
2014-12-01 33542, 2014
alastairp
hi
2014-12-01 33546, 2014
chirlu`
chirlu` has changed the topic to: Cooldown week | Agenda: osx signing key (alastair), lolo (bitmap)
2014-12-01 33519, 2014
alastairp
recent osx needs apps to be signed with a develpoer key before they open with no pain
2014-12-01 33531, 2014
ruaok
do we know what that entails?
2014-12-01 33531, 2014
alastairp
I belive this is already an issue with picard, and now we have the abz client too
2014-12-01 33541, 2014
alastairp
sign up as an apple developer
2014-12-01 33544, 2014
alastairp
$90/yaer
2014-12-01 33547, 2014
alastairp
sign
2014-12-01 33554, 2014
bitmap
we have several tickets asking for this for picard
2014-12-01 33503, 2014
ruaok
calendar year or 12 month period?
2014-12-01 33515, 2014
alastairp
good question. at a guess, 12 month period
2014-12-01 33531, 2014
alastairp
keeping in mind I have no mac-y things to test or do this on
2014-12-01 33535, 2014
ruaok
someone in the know, please gimme a ticket with instructions on what to do.
2014-12-01 33545, 2014
ruaok
bitmap: can you do that please?
2014-12-01 33552, 2014
alastairp
it would make sense if we could put metabrainz or similar in the sig
2014-12-01 33557, 2014
bitmap
sure, I'll look into it
2014-12-01 33500, 2014
alastairp
rather than a person. not sure if possible
2014-12-01 33500, 2014
ruaok
of course.
2014-12-01 33501, 2014
alastairp
cool
2014-12-01 33501, 2014
ruaok
thanks.
2014-12-01 33511, 2014
ruaok
bitmap: lolo.
2014-12-01 33523, 2014
chirlu`
chirlu` has changed the topic to: Cooldown week | Agenda: lolo (bitmap)
2014-12-01 33530, 2014
bitmap
so nikki and I think also reosarevok have reported that creating entities is noticeably slower than before the schema change
2014-12-01 33559, 2014
ruaok
thats odd.
2014-12-01 33500, 2014
reosarevok
Yup. At least from pop-ups
2014-12-01 33502, 2014
bitmap
we're not really sure how to figure out what's going on so I thought we could have another mbserver instance running somewhere with access to the live DB and also with CATALYST_DEBUG enabled
2014-12-01 33524, 2014
bitmap
which would tell us where the request is spending the most time
2014-12-01 33536, 2014
bitmap
and was wondering if it'd make sense to use lolo for that
2014-12-01 33536, 2014
ruaok
and lolo is the shortest route to that?
2014-12-01 33556, 2014
ianmcorvidae
for something that we can give more than just one person access to, probably
2014-12-01 33556, 2014
bitmap
well, I'm not sure where else would make sense
2014-12-01 33505, 2014
ianmcorvidae
and for the best emulation of a real setup
2014-12-01 33514, 2014
ruaok
I'll have lolo turned on. I'll ping ya when its on.
2014-12-01 33523, 2014
bitmap
cool
2014-12-01 33533, 2014
ruaok
do you want to work on it now (read: I call them) or soonish (I open a ticket)?
2014-12-01 33535, 2014
ianmcorvidae
I mean, you can do a tunnel and connect to the production DB from a devel machine if you have SSH access, but it's not exactly a *good* idea :P
2014-12-01 33507, 2014
bitmap
I don't think it's particularly urgent, so soonish sounds fine :)
2014-12-01 33513, 2014
ruaok
ok.
2014-12-01 33546, 2014
ruaok
mail sent.
2014-12-01 33550, 2014
ruaok
ok, that was it.
2014-12-01 33555, 2014
ruaok
anything else or get on with it?
2014-12-01 33543, 2014
ruaok
speaking on which...
2014-12-01 33550, 2014
ruaok
ianmcorvidae: do you still owe me a timesheet from last month?
2014-12-01 33558, 2014
ianmcorvidae
yeah, probably
2014-12-01 33508, 2014
ruaok
and of course this month. I received bitmap's and will send out an check
2014-12-01 33525, 2014
ruaok
ianmcorvidae: just a reminder. not urgent. both pending projects are more important.
2014-12-01 33557, 2014
ruaok
ok, lets close for today and get this show on the road. :)
2014-12-01 33503, 2014
ruaok
thanks everyone!
2014-12-01 33511, 2014
chirlu`
chirlu` has changed the topic to: Red-hot week | Agenda:
2014-12-01 33503, 2014
ruaok
well, stupid bank page doesn't load.
2014-12-01 33543, 2014
ruaok
bitmap: lolo is up.
2014-12-01 33502, 2014
bitmap
great
2014-12-01 33506, 2014
ruaok
please power it off again when you're done.
2014-12-01 33522, 2014
bitmap
alright
2014-12-01 33543, 2014
ijabz joined the channel
2014-12-01 33554, 2014
ianmcorvidae
bitmap: it's probably provisioned with chef like the other frontends, so you may need to amend server-configs or override it somehow
2014-12-01 33513, 2014
ianmcorvidae
(adding a new role for this sort of a debug server wouldn't be a bad idea anyway, so)
2014-12-01 33540, 2014
ianmcorvidae
(oh, maybe you can just add an override for that option in the node JSON directly. though it may need adding to chef-cookbooks' config file thing
2014-12-01 33511, 2014
ianmcorvidae
oh, no, catalyst_debug is already in it. cool
2014-12-01 33542, 2014
bitmap
ah, good
2014-12-01 33542, 2014
ianmcorvidae
so yeah, I think it should just be adding "musicbrainz-server" => { "dbdefs" => { "catalyst_debug" => true } } } to lolo.json
2014-12-01 33520, 2014
ianmcorvidae
but I don't quite know how the override behavior works so it might require copying roles/musicbrainz-server.rb, changing that one line, and then changing the role lolo.json uses
2014-12-01 33504, 2014
MBJenkins
* Ulrich Klauer: Remove "inc" directory from .gitignore
2014-12-01 33505, 2014
MBJenkins
* Ulrich Klauer: More improvements to INSTALL.md
2014-12-01 33506, 2014
MBJenkins
* Ulrich Klauer: Fix warning in InitDb.pl
2014-12-01 33507, 2014
MBJenkins
* Ulrich Klauer: Typographical fixes to README.md
2014-12-01 33522, 2014
bitmap
I guess they're all default attributes so what you pasted should override it
2014-12-01 33534, 2014
ianmcorvidae
yeah, I just don't know XD maybe it'd override the whole musicbrainz-server block, which would obviously not work very well, for example
2014-12-01 33532, 2014
bitmap rtfms
2014-12-01 33543, 2014
ianmcorvidae
heh
2014-12-01 33538, 2014
ianmcorvidae
better than me, I'd probably have just done the one that looked easier, deployed, and seen if it produced the right DBDefs :P
