#influxdb

it either somestring.some.other.string or 1h.somestring.some.other.string or 1d.somestring.some.other.string

so your serie can't match 2 regexp...

you're right then...

so if you have a catchall and it should be first... doesn't that mean that everything goes into the catchall and nothing will ever get into the specific ones?

first the catchall /.*/, then the /^1d\..*/ then /^1h\..*/

or does it mean, everything goes into the catchall AND into any other matching ones

that's the oposit !

it match the last (most specific) first, then go up until the catchall

the docs don't say that it "goes up"

you know you can try it by hand easily ?

when the docs tell me to order something in a specific way, i expect it to GO DOWN

the doc says to go from least to most specific...

just try it man !

no the docs just say that i should order it, the docs don't say what the database actually does with it

how? i can't see in which shard a series is saved.

well, wouldn't it be silly it they say to put the catall at top and everything goes it it without ever going to the others ?

yeap that would be silly UNLESS it will simply be saved in EVERY matching shard

(and that would sound like a massive waste of space to me)

mbrochh: that's why I don't expect it to work this way

though you're right you'll never know in which shard the serie lives

sure. i expect everything in the same way you do. but docs should be precise.

well, not now :)

Just installed the 8.3.. looking good so far

0.8.3 rather :)

big shards take a dogs age to load up on restart.. really painful

yeah, known issue, there's some discussion/tickets about it

yeah

huge shard too.. something like 30g.. I shouldn't let them be that large

Anyone have a vs-cassandra story?

ehhh, had to restart influxdb, it takes about 10~15seconds to open each shard

in 2 minutes, it opened about 12 shards

i think i have like 100 shards

wut

pickled: how long did it take for you?

most of my shards are 8MB big, i have a few of a few hundred MB, a few that are a few gig, and a few than are 12GB, those are the biggest

toddpersen: ^^

Dieterbe, I gave up after a few hours and deleted the shards

pickled: and then what?

then it started cleanly.. but i lost data

it was something like 3 or 4 hours chugging away at one shard

consuming no memory, and 100% of one core

really little disk access.. i didn't think it would finish tbh

hm ok it's up now for me

is there a way to make the "mean" aggregate function not round values?

we're storing cpu usage as a float, as in 0.1 = 10%, so values are often in the 0.0006 range

when using mean on that column it just returns 0's

hmm sounds like a bad way to store it

why not just store 10 instead of 0.1?

nlf, is 1.0 100% usage?

yes

weird

what does this?

iirc we're using a statsd collector and that's just how it sends it

diamond, specifically

i use diamond

but not via statsd

can't say i see that

maybe it's not via statsd, i'm not sure

let me check

that's broken either way it sounds

yeah, regardless it seems like mean shouldn't be rounding to 0

│ 1412367235000 │ 1 │ 95 │

│ 1412367225000 │ 1 │ 98 │

that's what I get on my idle series

i was wrong, this is coming from a java app's metrics

and apparently that's how java's metrics library sends data

yeah i've had experience with sigar

it's a joke

doesn't even write the cpu metrics close to what diamond or others report

and the best part is everyone blames me for running docker or Gentoo or they can't figure out why

i feel like you just described my experience with the jvm in general

computing likely

people would rather blame the person than take a look at the actual issue to see if it's valid

right

Damm: sometimes you should just lie about what you are using :)

Civil, remember when you lie to someone. You deny them the opportunity to review the situation honestly

I would be rather be honest and look bad; then lie and look worse

Damm: yeah, that's true, but sometimes you if you really sure that this bug has nothing to do with gentoo and docker you can pretend that you've tested it on something else.