Kubectl wait: Difference between revisions

From NovaOrdis Knowledge Base
Jump to navigation Jump to search
Line 45: Line 45:
The length of time to wait before giving up.  Zero means check once and don't wait, negative means wait for a week.
The length of time to wait before giving up.  Zero means check once and don't wait, negative means wait for a week.


{{Warn|The resource must '''exist''' for this option to be effective. If the resource is specified by name and does not exist, then the wait command exits immediately with an error message ("Error from server (NotFound): pods "..." not found"), irrespective of the timeout value. If the resource is specified by selector, the error message is "at least one resource must be specified to use a selector".}}
{{Warn|The resource must '''exist''' for this option to be effective. If the resource is specified by name and does not exist, then the wait command exits immediately with an error message ("Error from server (NotFound): pods "..." not found"), irrespective of the timeout value. If the resource is specified by selector, the error message is "at least one resource must be specified to use a selector" or "no matching resources found".}}


  ... --timeout=30s ...
  ... --timeout=30s ...

Revision as of 18:46, 17 March 2021

Internal

Overview

The command waits for a specific condition, as reported by the ".status.conditions" field on a pod state, on one or many pods. If more than one pod is specified, the condition must be seen in every pod. Alternatively, the command can wait for a specific pod to be deleted. The command exits with 0 if the resource is found and the conditions are met, or with a non-zero status if the resource is not found, or the condition is not met. Unless a specific timeout is provided with --timeout, the default timeout occurs in ? seconds.

Using kubectl wait with Other Resources then Pods

The documentation mentions that wait applies to "resources", not specifically pods. Can wait be applied for other things than pods? If yes, what is them meaning of "condition" in that context, as other resources may not have conditions? This discussion thread shows that the matter is not settled, at least at the time of writing: https://github.com/kubernetes/kubernetes/issues/80828

Wait for a Specific Condition

Wait on a Pod Specified by Name

kubectl wait --for=condition=Ready pod <pod-name>

Wait on a Pod Specified by Label

kubectl wait --for=condition=Ready -l color=blue

For more details on selector syntax see option -l, --selector below.

Wait for Deletion

Pod Deletion

Pod specified by name:

kubectl wait --for=delete pod <pod-name> --timeout=60s

Pod specified by selector:

kubectl wait --for=delete pod -l color=blue --timeout=60s

Namespace Deletion

kubectl wait --for=delete namespace <ns-name> --timeout=60s

Options

--timeout

The length of time to wait before giving up. Zero means check once and don't wait, negative means wait for a week.


The resource must exist for this option to be effective. If the resource is specified by name and does not exist, then the wait command exits immediately with an error message ("Error from server (NotFound): pods "..." not found"), irrespective of the timeout value. If the resource is specified by selector, the error message is "at least one resource must be specified to use a selector" or "no matching resources found".

... --timeout=30s ...

-l, --selector

Specifies the selector (label query) to use to identify the resource. The expression supports '=', '==', and '!='.

-l key1=value1,key2!=value2