Lifecycle

Start, stop, and delete for an environment mirror a workspace's, with one difference: a stop drains every service before it cuts.

Stop

Cuts a sync point of the disk, then removes every StatefulSet. State becomes stopped. Data is kept.

curl -sS -X POST https://dev.kloudlite.io/v1/environments/$ENV/stop -H "Authorization: Bearer $KL_TOKEN"

Start

Applies every service again on the node that holds the newest sync point.

curl -sS -X POST https://dev.kloudlite.io/v1/environments/$ENV/start -H "Authorization: Bearer $KL_TOKEN"

An environment running on a node that is down cannot be started elsewhere and cannot be cloned; it resumes when the node returns. A stopped one starts on any up-to-date node.

Delete

Removes the services and the disk. Pushed snapshots survive on a detached volume.

curl -sS -X DELETE https://dev.kloudlite.io/v1/environments/$ENV -H "Authorization: Bearer $KL_TOKEN"

Any intercept wishes on the environment go with it.

Restore

Put this environment back to one of its snapshots: its services and its data. Services drain, the disk is replaced, the snapshot's services come back. Data written since the snapshot is lost. restoring on the document reads Draining then Restoring while it runs, and restored_to names the snapshot afterwards.

curl -sS -X POST https://dev.kloudlite.io/v1/environments/$ENV/restore \  -H "Authorization: Bearer $KL_TOKEN" -H 'content-type: application/json' \  -d '{"snapshot_id": "push-env-2b8c-77aa"}'

A service is persistent when it has mounts. If a persistent service exists now and the snapshot does not have it, the restore would delete its data. The call answers 409 with lost, a list of {service, folders}, and writes nothing. Ask the person. Then repeat the call with discard naming those services. A name in discard that loses nothing is 422.

The 202 body lists snapshot_id, rolled_back, discarded, added and removed.

A snapshot that froze no services answers 422. Clone it into a new environment instead.