tarantool/tarantool-operator

applying roles to instances hang on kubernetes version 1.25+

vdovinmih opened this issue · 3 comments

vdovinmih commented

Expected behavior

roles are applied to instances without an infinite loop.

What do you want to achieve?
correct application of roles

Actual behavior

kubernetes versions 1.25 and higher (verified in versions 1.25, 1.26, 1.27) roles are applied to instances in infinite loop.

What is happening?
tarantool-operator applying roles to random instance in a loop. i see infinite logs in operator and cartridge instances

Steps to reproduce the behavior

  • use kubernetes versions 1.25

  • Add helm repository

    helm repo add tarantool https://tarantool.github.io/helm-charts/

  • Configure the Operator deployment (optional)

    helm show values tarantool/tarantool-operator > operator-values.yaml

    Edit operator-values.yaml file if needed, all parameters described in comments.

  • Install/Update operator (choose on of)

    helm upgrade --install tarantool-operator-ce tarantool/tarantool-operator [--values ./operator-values.yaml]

  • see logs

Environment

How are the pieces configured?

  • Tarantool Operator version: v1.0.0-rc3
  • Tarantool version: 2.11.1
  • Cartridge version: 2.8.2-1
  • Kubernetes version: 1.25
  • Helm chart configuration used: custom application with vshard-router and vshard-storage roles configured

Kubernetes version: 1.24.8 - works without a problem

d-enk commented

Tried with 1.25.15 and v1.28.7
Everything is okay
Need more info

vdovinmih commented

It is not reproducible in v1.28.7
Not sure about exact revisions 1.25, 1.26, 1.27.

I think we can close this

  • 👀1