#musicbrainz-devel

heh :)

annoying I can't finish this task today

you think that's bad? once it's done I then have to do a chmod

heh, fun. recursive, I'm sure

(since I imagine it's constant time if you're not recursing)

ianmcorvidae: that gist looks sane to me.

alright

and I can take the seach related bits, no biggie.

yeah, I figured everything not explicitly labeled as yours we'd split up

with more to me since I won't be sitting in a data center :)

You might consider warming the DB cache by running some queries such as SELECT * FROM artist on the big tables (or EXPLAIN ANALYZE SELECT ..., to discard the output immediately).

Makes the downtime a bit longer, but reduces 502s afterwards (hopefully).

100% 503s versus a few more 502s than usual, the former seems like it's blocking more legitimate usage :)

Because sequentially reading the whole table is faster than random access to the bits required for live pages.

might do that after things are back up though

Technically true, but psychologically people may be more annoyed by 502s after “everything is supposed to be back up” than by a few minutes more downtime.

The DB may also need a configuration change to actually use more RAM for caching.

ah, config is a good point

current shared_buffers is 10GB, I guess doubling that would make sense

:)

yes

looks good to me.

ok, time to move on. again. off to JS again.

ttfn

not javascript. san jose. not sure which is worse.

I suppose it might be a silly question since it's not exactly easy to notice, but is anyone else seeing any missing mb emails? there was a period of about 6 hours or something a few days ago where jira failed to send any emails (they didn't even appear in mb-bugs) and now someone has said that they didn't receive some edit note emails

and subscription emails

but are you receiving *all* of them? :P

hard to tell :)

exactly :/

my jira STYLE filter automatic email isn't coming every day, but I dunno if that's just because it didn't send it every day (no changes?) or what.

This should be ruaok's next 'do: http://www.dancehallusa.com/wp-content/uploads/...

sometimes I wish I had a fro

I wonder what my hair would do if I cut it short :P I haven't had it short since before puberty and my hair wasn't curly then

wtf is going on?

roobarb is being shit, apparently :P

ianmcorvidae: do you know where I need to change my pw to acces nagios.mb.org

?

I'm not sure where it can be changed, but I can show you where it's set, once I find it :)

no need to change it, really. I just wanted to see what else needs attention.

the rsync stuff apparently does.

there's one of them that has always failed to work

which should be disabled

not sure about others, the replication ones are at least *better* lately but not perfect

at least the search servers are looking fine now...

yeah, I kicked roobarb's jetty

voodoo.

thx

and restored them to equal weight

hi Gentlecat

So I think I have a MB slave server running ... are there unit tests for the entire system I can run?

enjoying SF?

not yet

just arrived there

heh, I just arrived in San Jose.

Alternatively, I can program my mediafile importer to do two lookups, one to my slave server and one to your master ones and compare? O.o

_Dave_: we don’t have anything to test a full live running setup.

ah ok :)

my super quick check is to do a search, then to load any random result.

I did that

that tests the web servers, DB and search servers.

if that is ok, the rest has a decent chance of being ok as well

there's a number of complex systems hooked together here

you should see the real site. :)

rube goldberg comes to mind. :)

lol

totally familiar with that, I've been an SA since 1980...seen enough of em

ianmcorvidae: so, looking at the nagios list, I can’t really tell which of the tests are still complaining. all thes ones that seem related to rsync have been failing for… nearly 2 years. :)

*someone

Submitting... I need to do less ABz stuff, I think.

ianmcorvidae: well, I see that all the replication packets are up to date.

I’ll just leave it be for now.

ruaok: well, at present nothing is actively in a warning or critical state, so that seems correct

ok. kewl.

> rob: Insert new memory

++ > rob: remove old memory

:)

I could just change it to "change memory" :P I'd assumed it was just adding in more sticks to empty slots, I guess

no, all slots are filled with 2GB sticks.

so, I need to yank those and replace them with 4GB sticks.

and yes, I’m just being pedantic. :)

ruaok: Well, your website sucks too.

(Actually, it is not too bad once NoScript is told your site isn't likely to burn down my apartment.)

(But it is a bit sucky to require JS. :p)

nikki: I noticed sometimes (but not recently) that JIRA emails were delayed, i.e. I could see updates in the activity stream, but the email notifications came hours later.

to their liking

gah.

bbiab