Updated system, installed, was happy it worked out of the box... and then looked into monitoring:
And suddenly I've got every service's CPU, mem and disk usage monitored every 10 seconds "for free".
A little bit of background: few months ago, when I was experimenting with cgroups and lxc I made a simple script to traverse cgroups and get statistics (cpu, mem, disk IO) and pipe it thru my monitoring system ( collectd -> riemann -> influxdb -> grafana ).
I was lukewarm about mandatory cgroups per service before ("meh, limiting is useful in 2 out 50 running services because most of them dont misbehave and package maintaners probably wont set them or set them wrong") but that is just awesome.
Now debugging any system slowness has become sooo much easier.
submitted by XANi_ to linux
[link] [6 comments]
Ei kommentteja:
Lähetä kommentti
Huomaa: vain tämän blogin jäsen voi lisätä kommentin.