#influxdb

so derivative is still horribly, horribly broken?

is there any fix planned? or should we just abandon it?

anyone here?

pauldix for example? ^^;

hi cnf

hey Dieterbe

cnf: here

pauldix: is derivative still doing weird things, or am I doing stupid things?

it was always flakey, but a few updates ago, it started to return either nothing or garbage

and my group by time() is 3 times my sample rate

select derivative(cpu_cumulative_usage), container_name from "stats" where container_name = 'hekad' group by time(15s) limit 1000; is my query atm

this is on data coming from cadvisor, btw

hmm, for me, it either returns nothing, or values several orders of magnitude off what is possible :/

cnf: there was a bug fix that went in that fixed one case, but broke others

cnf: we'll be backing that out and cutting a release today

oh, good

derivative is about 70% of my influxdb usecase

Good morning, all. I'm testing migration from 0.7.3 to 0.8.6. Ready to kick off the migration, but getting a 404 on the /cluster/migrate_data endpoint.

and I can't find any reference to migrate_data in the source (via GH search)

am I doing something wrong?

uh pauldix: you guys removed migration?

blalor: you should probably upgrade to a lower 0.8.* first that still has the migration stuff, and then update the point release if you want

yeah, that's what I'm doing

sort of flabbergasted that removed it, tho

s/that/that they/

I would expect more stability than that within a single release series (0.8.x)

i personally ran into a variety of issues when going from 0.8 rc3 to 0.8

0.8 feels like an extremely transitional version series.

makes me uneasy, actually.

yeah the migration was only meant to be there for a short time

I have odd feelings about that I admit

people straggle behind oddly

I was a pretty early adopter. I'd intended to jump to 0.8 when it came out, but clustering is still a red-headed step child. 0.7 has been solid for us.

It also seemed that 0.8 took a long time to become stable, so I held off.

We're getting ready to ramp up our monitoring tools that will use InfluxDB, so it makes sense to finally get serious about data retention and the like.

I am so excite that you guys are going the pure go route and embracing OOTB tooling.

as you land tag support I'll add an outputter to shh for metrics it collects. I've been hesitant to do that knowing tags will change things up a lot.

blalor, I don't blame you

blalor, I tend to burn myself by upgrading too fast