v0.9.1 - PR: #2031 - agent cpu usage explodes

[johnjcool]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

in contrast to 0.9.0 and 0.9.1 the fleet agent uses after the first DefaultResyncAgent much more cpu.

v0.9.1

v0.9.0

the main change between these two versions is this PR #2031. is it possible to revert this PR and fix the cpu usage problem first.__

Expected Behavior

small footprint at downstream clusters.

Steps To Reproduce

1. install fleet 0.9.1
2. deploy 40 BundleDeployments
3. wait until first DefaultResyncAgent
4. node cpu explodes

Environment

- Architecture: amd64
- Fleet Version: 0.9.1
- Cluster:
  - Provider: K3S
  - Options: 2
  - Kubernetes Version: 1.28.5

Logs

No response

Anything else?

No response

[kkaempf]

@raulcabello

[johnjcool]

also tested v0.8.3 and resource usage is fine

[thardeck]

@johnjcool many thanks for reporting and debugging the issue.

Could you verify if Fleet v0.9.2 fixes your issue?

[johnjcool]

v0.9.2-rc.1 fixes the issue:

[johnjcool]

many thanks for quick response time :-)