Skip to main content
Restarting ten pods, cordoning three nodes or cleaning up a batch of finished jobs takes one action, not ten. Pick the rows, choose what to do, and watch each object’s progress.
The pods list with several rows checked, and a bar at the bottom showing how many are selected and the actions for them.The pods list with several rows checked, and a bar at the bottom showing how many are selected and the actions for them.

Several pods picked, to act on them at once.

​
Pick rows

Selecting works in every list, and on the Workloads page across kinds. A bar at the bottom of the list says how many rows you picked, and offers the actions that apply to at least one of them.
Narrow the list first, with the filter (/), a health chip or a label selector, then press ⌘A to pick everything that’s left. See Search, filters and ⌘K.

​
What you can do

If an action applies to only some of the rows you picked, the dialog says how many it leaves out, like “Left out: 2 of the selected, which restart doesn’t apply to.” Those rows are left alone.

​
Run it

1

Choose the action

Choose it in the selection bar. The dialog lists every object it will change, and the equivalent kubectl commands: one line for each namespace and kind.
2

Confirm

Confirm, and the changes run, four at a time. Each object shows its own progress, and the reason if it fails.
3

Retry what failed

If some couldn’t be changed, the dialog says how many (“2 of 12 couldn’t be changed; see why below.”) and the button becomes Retry failed. It runs the action again, only for those.
When everything succeeds, a notification confirms it, like “Restarted 12 pods”.

​
Deleting many at once

Bulk delete removes each object the way a single delete does by default: anything an object owns is deleted after it. There’s no force option, and no choice of what happens to owned objects. For those, delete one at a time. See Actions by kind. Some deletes ask you to type what’s being done before they run:
  • Deleting Namespaces, Nodes, PersistentVolumes, PersistentVolumeClaims, StorageClasses or CustomResourceDefinitions.
  • Deleting anything in a cluster whose name looks like production. See Changing things safely.
You type the action and what it’s for, like delete namespaces, delete volume claims or, on the Workloads page, delete workloads.

​
Good to know

  • No undo. Changes made together can’t be undone from their notification. Each one is in the Activity log, with its command.
  • Permissions are checked by the cluster. Bulk actions don’t check your permissions before they start. Objects you can’t change fail with the cluster’s reason, and the rest go ahead.
  • Read-only clusters. In a read-only cluster, the actions in the selection bar are disabled.

Actions by kind

Everything you can do to one object, with all its options.

Undo and activity

Take changes back, and see everything you changed this session.