IRC log of musicbrainz-devel on 2010-08-05

Timestamps are in UTC.

00:55:33 [ruaok]

ruaok has joined #musicbrainz-devel

01:04:05 [ruaok]

so far, so good.

01:04:28 [ianmcorvidae]

awesome

01:04:47 [ianmcorvidae]

it seemed more responsive in my entirely unscientific and very brief tests :)

01:15:36 [ruaok]

the numbers behind the scenes back that up. :)

01:15:52 [ruaok]

hopefully I can get something useful done with the rest of my week

01:40:01 [jdamcd_]

jdamcd_ has joined #musicbrainz-devel

02:37:28 [pronik`]

pronik` has joined #musicbrainz-devel

04:11:53 [jensl]

jensl has joined #musicbrainz-devel

05:43:52 [djce]

djce has joined #musicbrainz-devel

06:03:14 [ruaok]

djce: hi

06:06:34 [dinog]

dinog has joined #musicbrainz-devel

06:16:07 [djce]

ruaok: hi

06:16:11 [djce]

so, went well?

06:16:12 [ruaok]

morning.

06:16:15 [ruaok]

yes.

06:16:22 [djce]

cool. load seems lower.

06:16:30 [djce]

down for ~ 90 min yes?

06:16:36 [ruaok]

someone suggested using cluster instead of vacuum full, but that took too long.

06:16:40 [ruaok]

sadly, yes.

06:16:53 [ruaok]

zero diskspace was recovered, but the load is low.

06:17:05 [djce]

hm .interesting.

06:17:08 [ruaok]

and we're not low on diskspace, so al is well.

06:17:11 [ruaok]

indeed.

06:17:13 [djce]

and you did all tables, presumably

06:17:35 [djce]

also yesterday I turned on stats collection,

06:17:40 [ruaok]

by hand, because I wanted to control the order in which they were done. but all were done and an analyze was also run.

06:17:50 [djce]

so now we have lots of interesting (non-null, non-zero) numbers in pg_stat*

06:17:58 [ruaok]

nice, good to know.

06:18:04 [ruaok]

but things look much much better.

06:18:14 [ruaok]

and it seems the daily reports ran without out of memory errors.

06:18:24 [djce]

cool. I hate to think when we last ran a proper vacuum

06:18:34 [djce]

you did "full" yes?

06:18:38 [ruaok]

yes.

06:18:48 [ruaok]

have we ever done a vacuum?

06:18:59 [djce]

ok. I did vacuum non-full on most (not all) yesterday, while the site was up

06:19:09 [ruaok]

its not something I consdidered, because how could we possibly go for 4 years without a vacuum?

06:19:13 [djce]

but chickened out of doing track & trackjoin, IIRC

06:19:14 [djce]

:-)

06:19:18 [ruaok]

or 3. whatever. a loooong time

06:19:31 [djce]

IIRC there's a vacuum command in admin/cron/daily

06:19:39 [djce]

I suspect it's commented out

06:19:46 [djce]

as being "too expensive" :-/ bad choice

06:19:55 [ruaok]

lol.

06:20:10 [ruaok]

massive fail. oh well. ;)

06:20:30 [ruaok]

but, it seems that our server is capable of handling the load.

06:20:35 [djce]

heh, no, it's not there at all.

06:20:39 [djce]

I mis-remember.

06:20:43 [ruaok]

but going to 9 and then replicating to a slave seems like a good idea.

06:20:57 [ruaok]

oh well.

06:21:09 [ruaok]

we should see about installing the auto-vacuum daemon.

06:21:15 [ruaok]

should I open a ticket for you?

06:21:29 [djce]

as I understand it, it is installed - just needs enabling

06:21:34 [djce]

and possibly configuring.

06:21:38 [djce]

yeah, shoot

06:21:43 [ruaok]

ah. ok. will do.

06:21:57 [djce]

and enabling == another postgres restart though, I think

06:22:07 [ruaok]

we can deal.

06:22:08 [djce]

or maybe an svc -h.

06:22:23 [ruaok]

I've gotten used to the process of taking the site up and down.

06:22:28 [ruaok]

MUUUUCH nicer than before.

06:22:30 [djce]

that's also good :-)

06:22:42 [ruaok]

though I think we ought to pizzaz our server down pages a little.

06:22:51 [ruaok]

maybe navap will give us a hand.

06:22:56 [djce]

sure. do we have a graphic designer?

06:23:03 [ruaok]

navap can help.

06:23:08 [ruaok]

can we serve an html page?

06:23:17 [djce]

I like the sleepy brainz, but yes I agree.

06:23:21 [djce]

yup, whatever you want

06:23:41 [ruaok]

I'm all for keeping the sleepy brainz.

06:23:43 [djce]

* djce imagines a 5xx error which is actually a streaming mp3

06:23:52 [djce]

nominate your song suggestions now! ;-)

06:23:57 [ruaok]

I would liike to add more info, such as a link to the blog that talks about what we're doing.

06:24:05 [ruaok]

* ruaok runs

06:24:08 [ruaok]

eeeek.

06:24:10 [djce]

ok

06:24:17 [ruaok]

badger badger badger bager MUUSHROOOM

06:24:22 [djce]

lol

06:24:42 [djce]

downtime downtime downtime downtime REBOOT REBOOT

06:25:06 [djce]

s/snake/fail!/ :-)

06:25:10 [ruaok]

lol

06:25:42 [ruaok]

our customers win on best downtime message:

06:25:43 [ruaok]

http://wisestartupblog.com/starting-up/grooveshark-goes-down-for-maintenance-cute-panda-in-its-place/1986/

06:26:15 [ruaok]

too bad the graphic is 404.

06:26:17 [ruaok]

its cute.

06:26:44 [djce]

didn't realise grooveshark was a customer. cool.

06:26:53 [djce]

they could do with good metadata :-)

06:26:59 [ruaok]

http://www.duncanleung.com/blog/2010/04/hungry-grooveshark-panda-pickles-nom-nom-nom/

06:27:10 [ruaok]

lol. :-)

06:28:00 [djce]

* djce likes the faded text in the "h2"

06:28:26 [ianmcorvidae]

hah

06:28:29 [ianmcorvidae]

I didn't even notice that

06:28:32 [ruaok]

we need to think up something cute so we can celebrate our failures too!

06:28:34 [ianmcorvidae]

clearly not paying enough attention :P

06:29:17 [djce]

ruaok: so, back to serious: we currently have nothing which regularly vacuums, yes? no cron job etc. hence the ticket.

06:29:30 [ruaok]

I would have to assume that, yes.

06:29:32 [djce]

ok

06:41:45 [ruaok]

djce: thanks for fixing MBH-41 so fast.

06:41:59 [djce]

<shrug> setting, restart, see what happens

06:42:02 [ruaok]

I supposed we could benefit from running ANALYZE for moderation_closed

06:42:06 [djce]

(and disabling mod_closed first :-)

06:42:07 [ruaok]

that doesn't take long to run.

06:42:28 [djce]

Maybe... but the table doesn't change much, other than growing

06:42:49 [djce]

so the stats won't really need updating will they?

06:42:53 [ruaok]

yep, and thus the tabel statistics change.

06:43:05 [ruaok]

over time they would.

06:43:16 [djce]

but in a meaningful way? Well I suppose eventually yes

06:43:27 [ruaok]

I don't see the impact as very great (since it doesn't scan the whole table, its statistically samples it) so I think its ok.

06:43:38 [ruaok]

yes, eventually.

06:43:48 [ruaok]

we could analyze it once a month I suppose to address that

06:44:00 [djce]

* djce nods

06:44:24 [ruaok]

how can we get feedback about the autovacuum deamon actually running and doing its job?

06:44:42 [ruaok]

I suppose I can periodically run the pgstattuple report again.

06:44:48 [ruaok]

its sitting on my screen in my office now.

06:44:52 [djce]

-bash-3.00$ sudo tail -F /service/pgsql-8.1/log/main/current | tai64nlocal

06:44:52 [ruaok]

I wonder how that has changed.

06:44:56 [djce]

2010-08-05 06:36:58.159657500 ERROR: invalid input syntax for integer: "zmhfzqwoxd"

06:44:56 [djce]

2010-08-05 06:38:31.418172500 LOG: autovacuum: processing database "template1"

06:44:56 [djce]

2010-08-05 06:39:32.704309500 LOG: autovacuum: processing database "postgres"

06:44:56 [djce]

2010-08-05 06:40:53.690692500 LOG: autovacuum: processing database "metabrainz_db2"

06:44:56 [djce]

2010-08-05 06:41:56.188074500 LOG: autovacuum: processing database "mbdb5"

06:45:09 [djce]

so it's running, at least.

06:45:33 [ruaok]

can we nagios that?

06:45:50 [djce]

somewhere pgsql records the last vacuum time for a table.

06:45:55 [djce]

That's what we should monitor.

06:46:00 [ruaok]

indeed!

06:46:01 [djce]

just need to find it...

06:46:13 [djce]

* djce goes to grab brekky, and start work

06:46:14 [djce]

biab

06:51:12 [djce]

* djce opens http://jira.musicbrainz.org/browse/MBH-42 "Nagios should check for non-vacuumed tables"

07:48:58 [dinog]

dinog has joined #musicbrainz-devel

08:19:22 [ijabz]

ijabz has joined #musicbrainz-devel

10:07:56 [djce]

djce has joined #musicbrainz-devel

11:14:36 [fatih]

fatih has joined #musicbrainz-devel

12:05:49 [jdamcd]

jdamcd has joined #musicbrainz-devel

12:05:54 [jdamcd]

jdamcd has left #musicbrainz-devel

12:06:03 [jdamcd]

jdamcd has joined #musicbrainz-devel

13:46:25 [ijabz_]

ijabz_ has joined #musicbrainz-devel

14:31:53 [kurtjx]

hello

14:31:57 [kurtjx]

not to make waves

14:32:10 [kurtjx]

but i've noticed that when you visit the page for a work

14:32:21 [kurtjx]

there aren't clear links to recordings of that work

14:33:00 [kurtjx]

i think w/o that, there's zero chance of editors of classical music accepting they should credit composers to works and performers to recordings

14:33:15 [kurtjx]

and we'll continue to have Bach with 3k recordings

14:36:05 [kurtjx]

our maybe i've got it all backwards :-)

14:36:28 [nikki]

that's just a presentation thing though, that's the case for a lot of the relationships in ngs at the moment

14:55:08 [kurtjx]

oh ok, so that's sorta on the todo list?

14:58:05 [kurtjx]

i should've known ;)

15:06:25 [warp]

kurtjx: our bug tracker is the todo list, if you cannot find a bug similar to this situation in jira... please just file a bug.

15:06:35 [warp]

kurtjx: that way you make sure we don't forget it :)

15:34:11 [Muz]

Muz has joined #musicbrainz-devel

15:45:47 [kurtjx]

thnx warp - i can't seem to find this in Jira - searching on "works"

15:53:20 [kurtjx]

ok it seems, on closer inspection, recordings of a work appear under relationships as "has performance"

15:53:34 [nikki]

* nikki nods

15:53:46 [nikki]

that's what I meant, I didn't realise you didn't know :/

15:54:11 [kurtjx]

perhaps that should appear in "overview" so chuckle heads like me don't get confused :)

15:54:36 [ianmcorvidae]

heh

15:54:40 [kurtjx]

is that a jira worthy feature request?

15:54:45 [nikki]

yes

15:55:27 [ianmcorvidae]

so, reading backlog, did we decide that new recommendation for classical will be "artist == performer, link to work, work's artist == composer"?

15:56:02 [kurtjx]

i don't think that was decided - it's what i think makes sense

15:56:06 [ianmcorvidae]

(still not completely accurate, of course, really you just can't do a unitary "artist" for classical, but as long as we're consistent :P)

15:56:09 [nikki]

and that's the point I was making :P most of the relationships in general are only shown on the relationships tab at the moment, there's already some code that shows artist-recording credits on recording and release pages, but there's plenty more relationships that need displaying better... but of course, the data's there, it's just needs displaying better

15:57:03 [kurtjx]

any way classical music is hard - let's just not listen to it :P

15:57:07 [ianmcorvidae]

kurtjx: fair enough; getting people to agree to the idea may be more iffy, but hopefully if the interface is good enough it could go through

15:57:11 [ianmcorvidae]

haha

15:57:50 [ianmcorvidae]

or we could display everything as composer/performer/work and drop this silly "artist"/"track name" thing ;)

15:58:12 [kurtjx]

yeah ianmcorvidae that's it - i think the NGS model inherently implies a composer is credited to a work

15:58:36 [nikki]

I'm still never sure how to credit performers in groups

15:58:49 [kurtjx]

but my understanding is that editors still want recordings directly credited to composers

15:58:53 [ianmcorvidae]

yeah; you don't compose a recording, you compose the work that gets recorded

15:59:12 [ianmcorvidae]

well, but with not much work we can at least display the composer for the relevant work on the recording page

15:59:16 [kurtjx]

for jira, if i'm talking about he current test server if affect version NGS Beta 2 ???

15:59:21 [nikki]

yes

15:59:28 [ianmcorvidae]

of course composer(s) and work(s) are more accurate :P

16:01:59 [kurtjx]

my first MB bug submission http://jira.musicbrainz.org/browse/MBS-804

16:02:12 [kurtjx]

not really "bug" but "improvement"

16:03:44 [kurtjx]

kurtjx has left #musicbrainz-devel

16:03:55 [kurtjx]

kurtjx has joined #musicbrainz-devel

16:31:29 [ruaok]

ruaok has joined #musicbrainz-devel

16:36:30 [kurtjx]

any idea what an entry with link_type.name='ROOT' means?

16:36:59 [ruaok]

thats the root of the link-type tree.

16:37:42 [kurtjx]

not sure i get it - there are several 'ROOT' entries

16:38:35 [kurtjx]

so each 'ROOT' is like a subject heading?

16:38:53 [ruaok]

* ruaok needs to look at it in the NGS context

16:39:39 [ruaok]

in NGS the link types have all be combined into one tree.

16:39:56 [kurtjx]

k

16:40:06 [ruaok]

so there is one ROOT element for each of the different link types.

16:40:06 [kurtjx]

so "ROOT" entries are artifacts of the old db

16:40:13 [ruaok]

not quite.

16:40:35 [ruaok]

a ROOT element signifiies the ROOT of a type tree for a given entity pairs.

16:40:47 [ruaok]

like artrist->artist, artist->url and so on.

16:43:10 [kurtjx]

ok so there's a ROOT for each possible combo of entities

16:43:38 [ruaok]

yep

17:18:24 [jdamcd]

ruaok: I've discovered a minor license issue, the ZXing integration classes are Apache license so not compatible with GPLv2

17:19:05 [ruaok]

wasn't there something about it being compat with GPLv3?

17:19:44 [jdamcd]

yeah, need to either switch to v3 or write my own code for what those classes do

17:19:58 [ruaok]

I think v3 would be prudent.

17:21:11 [ruaok]

yeah, that change is fine by me, if you're ok with it.

17:23:25 [jdamcd]

yeah, fine with me

17:52:31 [ruaok]

ruaok has joined #musicbrainz-devel

17:52:48 [ruaok]

feh. feh. feh.

17:52:59 [ruaok]

the load on the DB is high again. fuss.

17:58:22 [pronik`]

pronik` has joined #musicbrainz-devel

18:54:08 [alastairp]

alastairp has joined #musicbrainz-devel

19:46:42 [nikki]

nikki has joined #musicbrainz-devel

20:20:51 [ijabz]

ijabz has joined #musicbrainz-devel

20:38:53 [zazi]

zazi has joined #musicbrainz-devel

20:48:39 [jensl]

jensl has joined #musicbrainz-devel

20:53:15 [nikki]

ruaok: phone! :D

20:53:32 [ruaok]

YES!

20:53:33 [ianmcorvidae]

:D

20:53:34 [navap]

Awesome!

20:53:43 [ruaok]

navap's should arrive tomorrow.

20:53:44 [navap]

* navap can't wait for tomorrow

20:53:58 [ruaok]

nikki: did it make it in one piece without being damaged?

20:54:17 [nikki]

yep!

20:54:39 [ruaok]

your phone is already in NJ. :)

20:54:55 [ruaok]

nikki: excellent. I can't wait to hear if it works properly on your network.

20:55:08 [ruaok]

* ruaok wonders what nikki thought of the packing materials used.

20:55:14 [navap]

ruaok: Oh I've been keeping tabs on it ;)

20:55:25 [ruaok]

navap: :)

20:56:27 [navap]

I'm leaving for NJ tonight, by the time I get there tomorrow it should have already arrived at the house :)

20:56:30 [nikki]

ruaok: it does

20:56:44 [ruaok]

navap: perfect.

20:56:51 [ruaok]

nikki: awesome. 3g and all?

20:56:52 [nikki]

it didn't at first, I spent a couple of hours wondering wtf was going on

20:57:12 [nikki]

then I took the battery and sim card out for the 10th time, put it back in and all of a sudden it's all working perfectly

20:57:24 [ruaok]

odd.

20:57:34 [ruaok]

lets hope it doesn't do anything flaky.

20:57:34 [nikki]

yep, I get a 3g icon

20:57:38 [nikki]

* nikki nods

20:57:52 [nikki]

it could just be that I had trouble getting the cover off and back on, maybe I didn't push it back together properly

20:58:25 [ruaok]

I find that jabbing my finger into the speaker hole helps to slide the cover.

20:58:36 [ruaok]

are you going to use the skin I sent along?

20:58:44 [ruaok]

there are also some screen covers in there.

21:00:19 [nikki]

I think so

21:00:25 [nikki]

* nikki tries the skin thing

21:00:39 [ruaok]

its the best skin I've found so far.

21:00:48 [ruaok]

it makes the top button hard to push tho. :(

21:01:03 [ruaok]

if you dont plan on using the cover, maybe send it to navap?

21:04:36 [nikki]

* nikki quite likes it

21:04:44 [ruaok]

its nice, no?

21:04:51 [nikki]

yep :D

21:04:58 [nikki]

* nikki is having fun scanning barcodes now :P

21:05:03 [ruaok]

needs to be more orange, but I can live with it.

21:05:15 [ruaok]

have you installed the MB app already?

21:05:21 [nikki]

yep

21:05:27 [nikki]

that was one of the first things I did

21:05:32 [ruaok]

yay.

21:05:41 [ruaok]

don't forget to send chocolate, please!

21:05:54 [ruaok]

(not to me, this time)

21:06:12 [nikki]

I was really confused at first because it didn't find the cd I was scanning, then I realised it's probably using test and the barcode isn't on there

21:06:24 [ruaok]

correct.

21:06:37 [nikki]

who should I be sending it to? and do they have any preferences?

21:07:05 [ruaok]

didn

21:07:11 [ruaok]

't I send you an email on that?

21:07:29 [nikki]

hmm...

21:07:30 [nikki]

* nikki looks

21:07:43 [ruaok]

july 26.

21:09:11 [jdamcd]

yay, more people to break the Android app :)

21:09:42 [ruaok]

sadly is a monoculture of harware, but still more eyes.

21:09:54 [ianmcorvidae]

heh

21:10:25 [ruaok]

nikki, navap: I've asked for a clarification on what types of choco are desired.

21:10:28 [ianmcorvidae]

I might end up trying to put Android on my netbook; that would be different ;)

21:10:31 [nikki]

ok

21:13:21 [nikki]

* nikki wonders which german chocolate is good...

21:13:33 [nikki]

I always end up getting kuhflecken :P

21:13:53 [ianmcorvidae]

haha

21:13:59 [ianmcorvidae]

that stuff is admittedly good

21:14:30 [nikki]

I prefer the english name though... "happy cows" :D

21:14:39 [ianmcorvidae]

hah

21:18:57 [ruaok]

* ruaok drools

21:19:03 [ruaok]

that is one of my faves.

21:19:23 [ruaok]

but lets wait to hear more from google. then we can translate that into interesting choco orders.

21:19:30 [nikki]

:)

21:27:57 [ruaok]

navap: I'm sure you'v seen our "sleepy MB server" images that we display when MB is down?

21:28:24 [ruaok]

when I saw the grooveshark one with pickles the panda, I realized that we need to improve ours.

21:28:46 [ruaok]

more humor (read: celebrate your failures). and links to blog, etc...

21:35:46 [navap]

ruaok: Yeah I saw that panda image.

21:36:02 [ruaok]

we need something with more pizzaz. if you feel inspired. :)

21:36:09 [navap]

I'll see what I can do when I get back from NJ.

21:36:17 [ruaok]

k.

21:36:57 [navap]

We still haven't decided exactly when we're coming back, but I think it'll be somewhere around the 24th.

21:37:13 [ruaok]

quite some time.

21:37:32 [ruaok]

is it a vacation?

21:37:42 [ruaok]

* ruaok is not sure if going to NJ can be considered a vacation

21:45:09 [ruaok]

has anyone done a DB import recently?

21:45:18 [ruaok]

how big is the full DB on disk when imported?

21:47:15 [nikki]

mine's 9.8 gigs, but that's not a very recent import... from what I remember, the last time I did an import it was about 7 gigs

21:48:54 [nikki]

oh, and that's not including edits, I've never imported those

21:49:03 [ruaok]

with or without closed mods?

21:49:11 [nikki]

without

21:49:25 [ruaok]

I'll call the full import 10GB then.

21:49:31 [ruaok]

after a clean import, that is.

21:49:59 [ruaok]

I spent some time hanging out with the PG team a couple of weeks ago. its really handy to have a direct line to the team when I have problems. :)

21:50:11 [ruaok]

me being on the road does wonders for MB.

21:59:32 [navap]

ruaok: Yeah it's a vacation, a family reunion of sorts.

21:59:41 [ruaok]

ah, ok.

21:59:50 [navap]

I have grand parents and an aunt/uncle in NJ. I also had cousins there, but they moved to FL a few years back.

22:00:10 [ruaok]

* ruaok shakes his head at FL

22:00:21 [ruaok]

I lived there for two years. ick, ick, ICK.

22:00:28 [navap]

THe two cousins are coming up to NJ and my and my sister are going there as well. So it's going to be the four of us together under one roof, we haven't had that in quite a few years.

22:00:57 [navap]

Too many old people? Too many oranges? Too hot?

22:01:09 [ruaok]

old people yes.

22:01:16 [ruaok]

culturally dead.

22:01:30 [ruaok]

humid and fetid. ass backwards.

22:01:32 [navap]

heh

22:01:43 [ruaok]

car culture supreme.

22:51:04 [ijabz]

ijabz has joined #musicbrainz-devel

23:16:01 [ruaok]

ruaok has joined #musicbrainz-devel