You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Controller manager's --pod-eviction-timeout argument hasn't worked since Kubernetes v1.13. So unless you've added TaintBasedEvictions to your pods, they will take five minutes to failover when a node goes down. Setting global eviction timers is now done using the api server's
- --default-not-ready-toleration-seconds=
- --default-unreachable-toleration-seconds=
parameters. Since the kube_controller_pod_eviction_timeout variable is a string eg "5m", whereas the above expect an integer representing the number of seconds, should create two new variables: kube_apiserver_pod_eviction_not_ready_timeout and kube_apiserver_pod_eviction_unreachable_timeout.
The text was updated successfully, but these errors were encountered:
Controller manager's
--pod-eviction-timeout
argument hasn't worked since Kubernetes v1.13. So unless you've added TaintBasedEvictions to your pods, they will take five minutes to failover when a node goes down. Setting global eviction timers is now done using the api server's- --default-not-ready-toleration-seconds=
- --default-unreachable-toleration-seconds=
parameters. Since the
kube_controller_pod_eviction_timeout
variable is a string eg "5m", whereas the above expect an integer representing the number of seconds, should create two new variables:kube_apiserver_pod_eviction_not_ready_timeout
andkube_apiserver_pod_eviction_unreachable_timeout
.The text was updated successfully, but these errors were encountered: