20:40 PM
ianmcorvidae
you usually do updates via hobbes anyway, so do it in screen and let's do it as quick as we can :)
2013-05-15 13554, 2013
20:40 PM
warp
ok
2013-05-15 13503, 2013
20:41 PM
ianmcorvidae
asterix out
2013-05-15 13530, 2013
20:41 PM
warp
eep, that's quick.
2013-05-15 13552, 2013
20:41 PM
warp pulls master
2013-05-15 13553, 2013
20:41 PM
ianmcorvidae
heh, oops
2013-05-15 13500, 2013
20:42 PM
ianmcorvidae
well, it's out once you're ready :)
2013-05-15 13526, 2013
20:42 PM
warp fab productions asterix.
2013-05-15 13536, 2013
20:42 PM
andreypopp joined the channel
2013-05-15 13527, 2013
20:43 PM
warp
ianmcorvidae: done.
2013-05-15 13540, 2013
20:43 PM
ianmcorvidae
pingu out, asterix in
2013-05-15 13543, 2013
20:43 PM
warp
I see some warnings but I presume those are expected.
2013-05-15 13549, 2013
20:43 PM
ianmcorvidae
which ones?
2013-05-15 13501, 2013
20:44 PM
ocharles
ruaok: did you see one of our hypothesis on high load on totoro btw?
2013-05-15 13519, 2013
20:44 PM
warp
ianmcorvidae: [asterix] out: [warn]: duplicate MIME type "text/html" in /etc/nginx/sites-enabled/001-musicbrainz:16
2013-05-15 13524, 2013
20:44 PM
warp
ianmcorvidae: and the msgfmt one.
2013-05-15 13527, 2013
20:44 PM
ianmcorvidae
oh, yeah, whatever
2013-05-15 13532, 2013
20:44 PM
nikki
luks: upgrade done and replication up to date with no problems, thanks! :D
2013-05-15 13533, 2013
20:44 PM
ocharles
luks: ianmcorvidae caused that regression because I caused a different regression
2013-05-15 13536, 2013
20:44 PM
ocharles
:)
2013-05-15 13547, 2013
20:44 PM
ocharles
well, i guess it's not a regression
2013-05-15 13558, 2013
20:44 PM
ocharles
bug*
2013-05-15 13504, 2013
20:45 PM
warp
fatal
2013-05-15 13517, 2013
20:45 PM
ianmcorvidae
bugs, bugs everywhere
2013-05-15 13521, 2013
20:45 PM
ianmcorvidae
warp: hm?
2013-05-15 13555, 2013
20:45 PM
warp
ianmcorvidae: pingu failed presumably because it still though it was busy. seems to succeed now.
2013-05-15 13501, 2013
20:46 PM
warp
yes, done.
2013-05-15 13528, 2013
20:46 PM
ianmcorvidae
okay
2013-05-15 13534, 2013
20:46 PM
ianmcorvidae
ready to put pingu back in?
2013-05-15 13536, 2013
20:46 PM
warp
yes
2013-05-15 13552, 2013
20:46 PM
ianmcorvidae
pingu in, astro out
2013-05-15 13522, 2013
20:47 PM
warp
[astro] The server does NOT appear to be quiet!
2013-05-15 13538, 2013
20:47 PM
ocharles
documentation isn't load balanced yet is it?
2013-05-15 13541, 2013
20:47 PM
warp
now it is.
2013-05-15 13542, 2013
20:47 PM
ianmcorvidae
wikidocs will continue to happen on astro
2013-05-15 13555, 2013
20:47 PM
ianmcorvidae
until that PR gets done
2013-05-15 13557, 2013
20:47 PM
ianmcorvidae
(and reports)
2013-05-15 13503, 2013
20:48 PM
ianmcorvidae
but yeah, carry on :)
2013-05-15 13515, 2013
20:48 PM
warp
ianmcorvidae: astro done.
2013-05-15 13529, 2013
20:48 PM
ianmcorvidae
okay, everything back
2013-05-15 13547, 2013
20:48 PM
warp
yay, thanks
2013-05-15 13533, 2013
20:53 PM
MBJenkins
2013-05-15 13534, 2013
20:53 PM
MBJenkins
* warp: Merge branch 'mbs-6295'
2013-05-15 13552, 2013
20:54 PM
ocharles
heh, we can probably lose these jobs now
2013-05-15 13557, 2013
20:54 PM
ocharles deactivates them
2013-05-15 13557, 2013
20:54 PM
warp
:)
2013-05-15 13506, 2013
20:56 PM
MBJenkins
2013-05-15 13507, 2013
20:56 PM
MBJenkins
* warp: Merge branch 'mbs-6295'
2013-05-15 13511, 2013
21:01 PM
warp
ABORT ABORT
2013-05-15 13519, 2013
21:02 PM
ocharles
EPANIC
2013-05-15 13541, 2013
21:02 PM
warp
is the rest of the week a non-freeze week? :)
2013-05-15 13553, 2013
21:02 PM
warp
(what about next week?)
2013-05-15 13554, 2013
21:02 PM
ianmcorvidae
I think that's the notion
2013-05-15 13522, 2013
21:03 PM
ianmcorvidae
I think next week is not, though we may have to give way if something especially important is up
2013-05-15 13536, 2013
21:03 PM
ianmcorvidae
and we definitely need to ensure the GPG replication gets in, since we promised that :)
2013-05-15 13553, 2013
21:03 PM
warp
warp has changed the topic to: luke-warm week | DfOTS: 0 / Agenda: reviews
2013-05-15 13503, 2013
21:06 PM
ruaok returns
2013-05-15 13514, 2013
21:06 PM
ruaok
I did not catch the hypothesis on totoro...
2013-05-15 13554, 2013
21:06 PM
ianmcorvidae
short version is that ocharles and I both have pg9.2, which has some query planner improvements over 9.0 and 9.1
2013-05-15 13520, 2013
21:07 PM
ianmcorvidae
in the scrollback ollie found a couple that look like they may be especially relevant to release-event and area-related changes to some of our queries
2013-05-15 13538, 2013
21:09 PM
ocharles
ya
2013-05-15 13504, 2013
21:12 PM
ruaok
so, can we make a policy of only developing on pg version that we're actually supporting.
2013-05-15 13510, 2013
21:12 PM
ruaok
this isn't a new problem. yet it keeps happening.
2013-05-15 13517, 2013
21:12 PM
ruaok
+?
2013-05-15 13548, 2013
21:12 PM
nikki points to http://tickets.musicbrainz.org/browse/MBS-6161 which is still open :P
2013-05-15 13532, 2013
21:13 PM
ocharles
i do agree with that, but it's only a hypothesis that this is the problem
2013-05-15 13538, 2013
21:13 PM
ocharles
maybe 9.0 or 9.1 can do this rewriting
2013-05-15 13545, 2013
21:14 PM
ianmcorvidae
we require 9.0 at this point, or 9.1?
2013-05-15 13555, 2013
21:14 PM
ruaok
9.1
2013-05-15 13500, 2013
21:15 PM
ianmcorvidae is just going to commit the INSTALL changes for MBS-6161, that's stupid that we haven't changed it :P
2013-05-15 13501, 2013
21:15 PM
mb-chat-logger
2013-05-15 13509, 2013
21:15 PM
ruaok
maybe we can upgrade to 9.1 soon and mitigate this.
2013-05-15 13550, 2013
21:15 PM
Rockj
Q: what is cube.control extention for postgres?
2013-05-15 13553, 2013
21:15 PM
Rockj
extension
2013-05-15 13530, 2013
21:16 PM
ocharles
ruaok: do you have a migrated 9.1?
2013-05-15 13535, 2013
21:16 PM
ruaok
yes.
2013-05-15 13539, 2013
21:16 PM
ruaok
on sakura.
2013-05-15 13542, 2013
21:16 PM
ocharles
ok, then it would be good to test the query there
2013-05-15 13548, 2013
21:16 PM
ruaok
installed from packages, even.
2013-05-15 13552, 2013
21:16 PM
ianmcorvidae
which query were you using, ocharles?
2013-05-15 13555, 2013
21:16 PM
ruaok
feel free.
2013-05-15 13555, 2013
21:16 PM
ocharles
2013-05-15 13501, 2013
21:17 PM
ocharles
explain analyze that and let me know the plan
2013-05-15 13539, 2013
21:18 PM
ianmcorvidae
0.855ms for that
2013-05-15 13541, 2013
21:18 PM
ianmcorvidae
that bodes well
2013-05-15 13549, 2013
21:18 PM
ocharles
does anyone have 9.0
2013-05-15 13551, 2013
21:18 PM
ocharles
that's not totoro :p
2013-05-15 13554, 2013
21:18 PM
ianmcorvidae
rika does
2013-05-15 13535, 2013
21:19 PM
ocharles
sakura presumably just has a bunch of index lookups in a nested loop?
2013-05-15 13540, 2013
21:19 PM
ianmcorvidae
2013-05-15 13551, 2013
21:19 PM
ianmcorvidae
lemme go run it on rika
2013-05-15 13500, 2013
21:20 PM
ianmcorvidae
(that one's sakura, on 9.1)
2013-05-15 13542, 2013
21:20 PM
ocharles
yep, sakura matches my machine
2013-05-15 13545, 2013
21:20 PM
ruaok
ocharles: there is a from source install of 9.0 complete with db on sakura.
2013-05-15 13554, 2013
21:20 PM
ruaok
sakura has complete 9.0 and 9.1 installs.
2013-05-15 13556, 2013
21:20 PM
ianmcorvidae
rika, 9.0
2013-05-15 13558, 2013
21:20 PM
ianmcorvidae
2013-05-15 13513, 2013
21:21 PM
ruaok
one from package and one from source in /usr/local/postgres-9.0.13
2013-05-15 13514, 2013
21:21 PM
ocharles
yea, rika matches totoro
2013-05-15 13527, 2013
21:21 PM
ianmcorvidae
this suggests that an upgrade to 9.1 will solve a lot of these problems?
2013-05-15 13500, 2013
21:22 PM
Rockj
regarding my question: sicbrainz-server/admin/sql/Extensions.sql:4: ERROR: could not open extension control file "/usr/share/postgresql/9.1/extension/cube.control": No such file or directory
2013-05-15 13502, 2013
21:22 PM
ocharles
we have several 9.1 machines with the right plan and no 9.0 machines with it
2013-05-15 13516, 2013
21:22 PM
ocharles
so yes, 9.1 does agree with the hypothesis
2013-05-15 13522, 2013
21:22 PM
ocharles
Rockj: did you install postgresql-contrib?
2013-05-15 13544, 2013
21:22 PM
Rockj
oh of course
2013-05-15 13546, 2013
21:22 PM
Rockj
doh
2013-05-15 13522, 2013
21:23 PM
Rockj
there we go, importing :-)
2013-05-15 13527, 2013
21:23 PM
Rockj
on new server
2013-05-15 13556, 2013
21:25 PM
MBJenkins
2013-05-15 13557, 2013
21:25 PM
MBJenkins
Ian McEwen: We require postgresl 9.1.
2013-05-15 13523, 2013
21:26 PM
ianmcorvidae
heh, I spellz gud
2013-05-15 13524, 2013
21:26 PM
reosarevok
woot, that didn't break :p
2013-05-15 13525, 2013
21:26 PM
ianmcorvidae
anyway
2013-05-15 13513, 2013
21:27 PM
warp
so we sudo apt dist-upgrade totoro until we're at 12:04 LTS and then sudo apt-get install postgres? :)
2013-05-15 13516, 2013
21:27 PM
ocharles
master built in 8 minutes?
2013-05-15 13520, 2013
21:27 PM
ocharles
that doesn't seem right
2013-05-15 13501, 2013
21:28 PM
ianmcorvidae
er
2013-05-15 13502, 2013
21:28 PM
ianmcorvidae
that's true
2013-05-15 13516, 2013
21:28 PM
ianmcorvidae
it's almost certainly not related to that patch, since that was a change to INSTALL.md, but
2013-05-15 13528, 2013
21:28 PM
ocharles
yet the log does look like it ran everything...
2013-05-15 13530, 2013
21:28 PM
ocharles
i am tres confused
2013-05-15 13539, 2013
21:28 PM
ianmcorvidae
is the html5 validator hooked up?
2013-05-15 13548, 2013
21:28 PM
ianmcorvidae
ok 4 # skip No HTML_VALIDATOR configured, skip html validation
2013-05-15 13552, 2013
21:28 PM
Jozo
(I tried read source) but is it important split every entitity to own file on geordi? (used discogs example... )
2013-05-15 13500, 2013
21:29 PM
warp
ianmcorvidae: master doesn't do html5 validation.
2013-05-15 13509, 2013
21:29 PM
warp
ianmcorvidae: only the beta build on jenkins is configured to validate HTML
2013-05-15 13516, 2013
21:29 PM
ianmcorvidae
Jozo: separate folders, they should be named by entity
2013-05-15 13520, 2013
21:29 PM
ianmcorvidae
er, named by ID
2013-05-15 13541, 2013
21:29 PM
ianmcorvidae
at present it supports only having one base type of entity, and everything else is done as a 'subitem'
2013-05-15 13509, 2013
21:30 PM
Jozo
ianmcorvidae: Oh, thanks
2013-05-15 13512, 2013
21:30 PM
ianmcorvidae
(generally, that's a release)
2013-05-15 13535, 2013
21:31 PM
warp
ocharles: yep, I also don't see anything obviously wrong in the logs.
2013-05-15 13520, 2013
21:32 PM
ocharles
2013-05-15 13540, 2013
21:32 PM
Jozo
ianmcorvidae: When works is covered on geordi?
2013-05-15 13502, 2013
21:33 PM
warp
ocharles: yes, so we broke it after build #423.
2013-05-15 13512, 2013
21:33 PM
warp
(or fixed it, depending on wtf is going on)
2013-05-15 13523, 2013
21:33 PM
ianmcorvidae
2013-05-15 13539, 2013
21:33 PM
ocharles
Oh
2013-05-15 13540, 2013
21:33 PM
ianmcorvidae
Jozo: it can be works now, but we haven't really tried yet
2013-05-15 13545, 2013
21:33 PM
ianmcorvidae
Jozo: what're you trying to do?
2013-05-15 13547, 2013
21:33 PM
ocharles
I did change the web service tests to not do TRUNCATE
2013-05-15 13559, 2013
21:33 PM
warp
2013-05-15 13516, 2013
21:34 PM
ocharles
that's mad.
2013-05-15 13519, 2013
21:34 PM
ocharles gives himself a raise
2013-05-15 13538, 2013
21:34 PM
warp
lol
2013-05-15 13510, 2013
21:35 PM
warp
I'm still waiting for that postgresql upgrade which will make TRUNCATEs fast.
2013-05-15 13528, 2013
21:35 PM
ocharles
that'd be nice
2013-05-15 13531, 2013
21:35 PM
ianmcorvidae
"TRUNCATE quickly removes all rows from a set of tables. It has the same effect as an unqualified DELETE on each table, but since it does not actually scan the tables it is faster." apparently completely incorrect
2013-05-15 13540, 2013
21:35 PM
ianmcorvidae
(quoted from current postgres docs)
2013-05-15 13546, 2013
21:35 PM
ianmcorvidae
why is it that truncate is slower?
