deploy command


Usage: kluctl deploy [flags]

Deploys a target to the corresponding cluster This command will also output a diff between the initial state and the state after deployment. The format of this diff is the same as for the ‘diff’ command. It will also output a list of prunable objects (without actually deleting them).


The following sets of arguments are available:

  1. project arguments
  2. image arguments
  3. inclusion/exclusion arguments

In addition, the following arguments are available:

Misc arguments:
  Command specific arguments.

      --abort-on-error               Abort deploying when an error occurs instead of trying the remaining deployments
      --dry-run                      Performs all kubernetes API calls in dry-run mode.
      --force-apply                  Force conflict resolution when applying. See documentation for details
      --force-replace-on-error       Same as --replace-on-error, but also try to delete and re-create objects. See
                                     documentation for more details.
      --no-wait                      Don't wait for objects readiness'
  -o, --output-format stringArray    Specify output format and target file, in the format 'format=path'. Format
                                     can either be 'text' or 'yaml'. Can be specified multiple times. The actual
                                     format for yaml is currently not documented and subject to change.
      --readiness-timeout duration   Maximum time to wait for object readiness. The timeout is meant per-object.
                                     Timeouts are in the duration format (1s, 1m, 1h, ...). If not specified, a
                                     default timeout of 5m is used. (default 5m0s)
      --render-output-dir string     Specifies the target directory to render the project into. If omitted, a
                                     temporary directory is used.
      --replace-on-error             When patching an object fails, try to replace it. See documentation for more
  -y, --yes                          Suppresses 'Are you sure?' questions and proceeds as if you would answer 'yes'.


kluctl implements deployments via server-side apply and a custom automatic conflict resolution algorithm. This algurithm is an automatic implementation of the “Don’t overwrite value, give up management claim” method. It should work in most cases, but might still fail. In case of such failure, you can use --force-apply to use the “Overwrite value, become sole manager” strategy instead.

Please note that this is a risky operation which might overwrite fields which were initially managed by kluctl but were then overtaken by other managers (e.g. by operators). Always use this option with caution and perform a dry-run before to ensure nothing unexpected gets overwritten.


In some situations, patching Kubernetes objects might fail for different reasons. In such cases, you can try --replace-on-error to instruct kluctl to retry with an update operation.

Please note that this will cause all fields to be overwritten, even if owned by other field managers.


This flag will cause the same replacement attempt on failure as with --replace-on-error. In addition, it will fallback to a delete+recreate operation in case the replace also fails.

Please note that this is a potentially risky operation, especially when an object carries some kind of important state.


kluctl does not abort a command when an individual object fails can not be updated. It collects all errors and warnings and outputs them instead. This option modifies the behaviour to immediately abort the command.

Last modified September 27, 2022: docs: Update commands (789c019)