Project

General

Profile

Enhancement #467

Enhancement #290: More sup indicators

Per-service accounting

Added by Marc Dequènes over 4 years ago. Updated almost 4 years ago.

Status:
Resolved
Priority:
Normal
Category:
Service :: Supervision
Start date:
2015-07-18
Due date:
% Done:

100%

Estimated time:
Patch Available:
Confirmed:
No
Branch:
Entity:
DuckCorp
Security:
Help Needed:
Yes

Description

Now that all machines are under systemd, it is easier, as all services are in a specific cgroup. Accounting should be activated and an LLD/template built.

The following per-service indicators would be nice:
  • CPU
  • Memory
  • I/Os
  • Network

History

#1

Updated by Marc Dequènes over 4 years ago

  • Status changed from New to In Progress
  • Assignee set to Marc Dequènes
  • % Done changed from 0 to 10
Activation of accounting has been done on all machines but:
  • only Orfeo and Toushirou were rebooted afterwards, but that's enough for creating the necessary config
  • memory accounting in the kernel is activated according to dmesg and the systemd code seems ok but /sys/fs/cgroup/memory is missing
  • network accounting does not seem available

The LLD discovery script with a template using it and creating CPU items and graphs was created. I was not sure of the formula so i tried to mimic what systemd-cgtop does. I am not confident about the result.

#2

Updated by Marc Dequènes over 4 years ago

  • Help Needed set to Yes
#3

Updated by Marc Dequènes about 4 years ago

  • % Done changed from 10 to 20

I fixed the gathering and calculation of numbers to return integers (when appropriate) instead of scientific notation, because Zabbix cannot cope with this format.

#4

Updated by Marc Dequènes almost 4 years ago

  • Status changed from In Progress to Resolved
  • % Done changed from 20 to 100

It seems to work nicely.

Network accounting is not available in systemd yet.

All machines may not have all information due to pending reboot.

Also available in: Atom PDF