Applying Changes

Policy changes apply when a terminal is newly provisioned. They do not rewrite a container that is already running.

This affects image changes, env var changes, resource changes, storage changes, and reset settings.

You do not need to restart the orchestrator for policy image or env changes. Save the policy, then refresh the affected user terminals. Restart the orchestrator only when you change the orchestrator deployment itself.

Refresh Terminals

After changing a policy, refresh matching terminals:

curl -X POST http://terminals-orchestrator:3000/api/v1/terminals/refresh \
  -H "Authorization: Bearer $TERMINALS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "policy_id": "data-science",
    "only_idle": true
  }'

only_idle defaults to true, so active users are not interrupted. Set it to false only when you intentionally want to stop matching running terminals immediately.

To refresh one user's terminals:

{
  "user_id": "user-123",
  "policy_id": "data-science"
}

That covers every terminal the user has under that policy. On a connection using Terminal Contexts they have more than one, so add context_id to refresh just one of them:

{
  "user_id": "user-123",
  "policy_id": "data-science",
  "context_id": "chat:0f3c1a2b-7d55-4a19-9c0e-2b8f6d4e1a37"
}

Use default for the shared terminal, chat:<chat-id> for one chat's terminal and automation:<automation-id> for one automation's terminal.

To also wipe persisted terminal files during refresh:

{
  "policy_id": "training-lab",
  "only_idle": true,
  "reset": true
}