All checks were successful
test-build / guarddog (push) Successful in 2m14s
test-build / build (push) Successful in 2m17s
release-image / guarddog (push) Successful in 45s
release-image / build (push) Successful in 1m42s
release-image / semantic-release (push) Successful in 48s
release-image / release-harbor (push) Successful in 8m54s
release-image / release-gitea (push) Successful in 3m20s
renovate / renovate (push) Successful in 1m3s
28 lines
1.1 KiB
Plaintext
28 lines
1.1 KiB
Plaintext
---
|
|
title: Descheduler
|
|
description: Descheduler, based on its policy, finds pods that can be moved and evicts them.
|
|
hero:
|
|
tagline: Descheduler, based on its policy, finds pods that can be moved and evicts them.
|
|
image:
|
|
file: https://cdn.jsdelivr.net/gh/selfhst/icons@main/webp/kubernetes.webp
|
|
actions:
|
|
- text: Source
|
|
link: https://github.com/kubernetes-sigs/descheduler
|
|
icon: right-arrow
|
|
- text: Deployment Chart
|
|
link: https://gitea.alexlebens.dev/alexlebens/infrastructure/src/branch/main/clusters/cl01tl/helm/descheduler
|
|
icon: right-arrow
|
|
---
|
|
|
|
import { Aside } from '@astrojs/starlight/components';
|
|
|
|
<Aside type="tip">Has an internal [Grafana Dashboard](https://grafana.alexlebens.net/goto/afit9jt9odji8a?orgId=1).</Aside>
|
|
|
|
# Purpose
|
|
|
|
Automated service that balances pods across the nodes according to configuration.
|
|
|
|
# Notes
|
|
|
|
Mostly used to balance number of pods. Current limit is set to 200 which is nearly twice the defaults, but this can sometimes be met during reboots during upgrades as pods are evicted. Then this will balance out one all nodes are healthy again.
|