Skip to content

Manage an instance

Every instance has a Manage menu, in the list and on its page. It offers only what the instance's state allows. Each action is an operation: the console accepts it at once and the location carries it out.

The Manage menu of an active instance
The Manage menu of an active instance

Before you start

You need a role of project admin or project member. A locked instance offers only Unlock.

What each state allows

State Actions
Active Stop, Reboot, Pause, Suspend, Shelve, Resize, Create image, Rebuild, Rescue, Lock
Stopped Start, Shelve, Resize, Create image, Rebuild, Rescue, Lock
Paused Resume
Suspended Resume
Shelved Unshelve
Rescued Unrescue
Error Reboot, Rebuild, Rescue
Locked Unlock, and nothing else

Rename and Delete are always offered, except while the instance is changing.

What each action does

Action Effect API
Stop / Start Shuts it down and starts it again. Disks, addresses, settings and its place on its host stay. :stop, :start
Reboot Restarts it. Programs are interrupted and connections drop. :reboot (type: soft or hard)
Pause / Resume Freezes it in memory; nothing runs until you resume. :pause, :unpause
Suspend / Resume Saves its memory to disk and stops it. :suspend, :resume
Shelve / Unshelve Shuts it down and gives up its place on its host. Unshelving takes longer than a start. :shelve, :unshelve
Resize Moves it to another size; see below. :resize
Rebuild Replaces its system disk with a fresh copy of an image; see below. :rebuild
Rescue / Unrescue Restarts it from a fresh copy of its image, with its own disk attached for repair. :rescue, :unrescue
Lock / Unlock Stops anyone changing or deleting it until it is unlocked. :lock, :unlock
Create image Saves its system disk as one of the project's images. :create-image
Rename Changes its name, here and in the location's tools. PATCH

Every action is a POST to …/servers/{server_id}:<action> and answers 202 Accepted:

curl -sS -X POST "$NUBESTACK_API/v1/orgs/$ORG/projects/$PROJECT/servers/srv_01jbweb1:stop" \
  -H "Authorization: Bearer $NUBESTACK_API_KEY" \
  -H "Idempotency-Key: $(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{}'

Resize

Resize moves an instance to another size. It restarts on the new size and comes back as it was, running or stopped.

  1. Choose Manage → Resize.
  2. Pick the New size from the sizes the location offers.
  3. Choose Resize and follow the operation.
The resize dialog: the current size and the sizes the location offers
The resize dialog: the current size and the sizes the location offers
curl -sS -X POST "$NUBESTACK_API/v1/orgs/$ORG/projects/$PROJECT/servers/srv_01jbweb1:resize" \
  -H "Authorization: Bearer $NUBESTACK_API_KEY" \
  -H "Idempotency-Key: $(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{"offering": "large"}'

Quota and capacity are checked first. A size with a smaller root disk is refused. There is no trial period: to go back, resize again.

Rebuild

Rebuild replaces the system disk with a fresh copy of an image you choose. Everything written to that disk is lost. Its addresses, volumes, security groups and SSH key stay.

Choose Manage → Rebuild, pick the Image and choose Rebuild instance.

curl -sS -X POST "$NUBESTACK_API/v1/orgs/$ORG/projects/$PROJECT/servers/srv_01jbweb1:rebuild" \
  -H "Authorization: Bearer $NUBESTACK_API_KEY" \
  -H "Idempotency-Key: $(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{"image": "ubuntu-24-04"}'

Rescue a broken instance

When an instance no longer starts, Rescue starts it from a fresh copy of its image, with its own disk attached as an extra disk. Log in, mount the disk, fix it, then choose Unrescue to start from its own disk again. The console log often shows why it would not start.

Tags and metadata

  • Tags tab: key: value labels for filtering lists and grouping costs. See Tags.
  • Metadata tab: choose Change metadata to set or remove keys the instance reads from inside. A key is 1 to 64 letters, digits, - or _; a value at most 255 characters. Keys starting with nexus are reserved. The console cannot list the keys an instance already has; the location's own tools show them.

Good practice

  • Stop keeps an instance's place on its host and starts quickly. Shelve frees the place; use it for an instance you will not need for a while.
  • A stopped instance still counts for its disks and public IPs. See Usage and costs.
  • Lock production instances so nobody stops or deletes them by mistake.
  • To stop and start instances on a timetable, use a power schedule.