some memory profiling, mostly, though it may be within postgres that the problem lies
in fact, the weekly ones might be faster if the bottleneck is the download :)
ruaok
ianmcorvidae: I think it might be doing too many transactions.
one for each XID might be the "correct" thing to do, but also slows things down.
I wonder if we can safely batch more changes.
ianmcorvidae
I'm not sure
Lonewlf
ruaok : something like, "I see you are running a replication from more than a week back, would you like to apply the full dump? (This is much faster!!!)"
and script it :)
ruaok
Lonewlf: its not quite that easy.
for people who are running in mission critical environments, the replication is easier than swapping in a new DB.
Lonewlf
ruaok : no, it never is, but that's what you'd like it work like :P
ruaok : indeed
ianmcorvidae
oh, hey, I have access to the musicbrainz-server repo, didn't know that
did cd musicbrainz-server then cd admin then ./psql READWRITE
ianmcorvidae
mapti89: carton install Statistics::Basic
mapti89: you may be seeing a pattern here :)
Lonewlf
and now I'm stumped on the syntax fro dropping the db
ruaok
Lonewlf: you need to drop the DB as user postgres, outside of psql
ianmcorvidae
Lonewlf: sudo -u postgres dropdb musicbrainz_db
I think that ruaok missed the '-u'
ruaok
oh, I normally do sudo su postgres
I didn't even know about the -u flag. :)
Lonewlf
oh right ok
ianmcorvidae
ah :)
ruaok: how kosher is it for me to poke at the frontend stuff for test? our deployment seems to have made some strange choices re: protocol support and ciphers and things
ruaok
ianmcorvidae: poke away!
you've got the keys to the kingdom and you have sense in your head. poke it!
ianmcorvidae
k
just still wary about touching frontend stuff, given the potential to accidentally burn the world :)
ruaok
you mean nginx on carl/lenny?
nginx -t
is your friend. ;)
ianmcorvidae
well, yes
even so :)
ruaok
do you need some guideance on what to change?
ianmcorvidae
nah, I think for test I'm set
ruaok
k.
we really need ingestr. like yesterday.
ianmcorvidae
disabling insecure SSL versions, improving the lists of ciphers, etc.
ruaok
more people want to send us data. and we have no place to put it. :(
hopefully we can make some progress on this on Oct 22nd.
alastairp
ahhh. 22nf
that's so close
ruaok
thats our day at the archive.
not the MB summit, alastairp
alastairp
oh. so it is
sigh. so much on
ianmcorvidae
ruaok: actually, I do have one question: how do I make nginx reload its config? :)
ruaok
cd /usr/local/nginx
ianmcorvidae
I realize we're not using plain sysvinit for that, so
ruaok
svc -h .
we're using daemontools
ianmcorvidae
yeah, I've never used it before, hence the question :)
thanks
jdamcd joined the channel
hawke_1 joined the channel
hmph
ijabz joined the channel
Lonewlf
so, another question... having blown away my database so that I can run the import, did I shoot myself in the foot in some regard by only getting the 3 files suggested? should I have gotten all of them?
ijabz
plaintext: hi the inc analyis looks good but Im surprised there doesn't seem to be a section for releases
Lonewlf
I only goet mdump, the derived and the editor
nikki
those are the only ones you need
ijabz
i think this is the entity I use most, and I suspect others do as well
nikki
although the others are pretty small, so I normally import those anyway
ianmcorvidae
ijabz: the 'Summary' section is actually releases, I think we determined
ijabz: with the wrong label obviously, but
plaintext
ijabz: I found the bug that causes this, and I'm working on it :)
btw, I think Summary is just Summary
hawke joined the channel
ijabz
Is summary just the top ones over any entities, now I look at it some clearly seem to be for release and others clearly not
plaintext
yes it's the top for every entity
ianmcorvidae
just releases are the most common, I see
bah c'mon nginx why won't you notice my config changes
theeeere we go
Lonewlf
hey thanks guys, I'll pop in again if I blow myself up again
ianmcorvidae
other than our cert being self-signed we're doing pretty well on ssllabs' test now
ruaok
ianmcorvidae: why not use the new wildcard cert.
the deets on are carl.
ianmcorvidae
ruaok: was working on other things first is all, that's next :)
ruaok
k
mapti89
Compilation failed in require at app.psgi line 21.
BEGIN failed--compilation aborted at app.psgi line 21.
ruaok
carton install Catalyst::Utils
ianmcorvidae
catalyst::utils looks like it's already there
it's failing in that file after all
not sure what's going on there
mapti89
You have Catalyst::Utils (undef)
ianmcorvidae
whoo, test now using the real cert
huh, apparently the wildcard has plain musicbrainz.org as a SAN
ijabz joined the channel
mapti89
someone got an idea?
plaintext
yay, bug fixed
it feels nice when fixing a bug results in much cleaner code
ianmcorvidae
haha, yeah
I love when I can delete huge swaths of code as part of a patch :D
ruaok
ok, I'm off to have lunch with my stalker.
plaintext
heheh
ianmcorvidae
ocharles, warp: could one of you pull in mbs-5301-ssl-gravatars and deploy it to test for me?
why is my firefox still telling me the cert is insecure
answer seems to be "my firefox is poorly configured"
mapti89: I'd try running plain 'carton install' and seeing if it pulls anything in
mapti89: looking a bit deeper that error is being thrown when it can't load *something*, but unfortunately it's not printing what it's failing to load that I can see
jdamcd joined the channel
hawke joined the channel
mapti89
ianmcorvidae: after running carton install, did i have to run: "carton install --deployment"?
ianmcorvidae
they do largely the same thing; --deployment looks for specific versions while without it doesn't
so no, you should just need one or the other
see if there were any errors, though, of course :)
mapti89
yes ;-)
the command install a lot of things that are missing!