oz0 kill
For: Operators and developers looking up a command.
Kill switches stop one agent (or flow), one tool or one tenant on every edge and node within seconds: no new jobs start, model calls are refused, and running jobs are paused (--mode park, the default; they resume when the switch is released) or cancelled (--mode cancel). A tool's switch makes every node refuse its calls. Setting and releasing a switch is in the audit log, with who and why.
A switch bumps the tenant's authority epoch: every edge refuses the calls of the jobs it reaches, a child another agent started of a stopped agent included, keyed by each call's binding; an edge stops exactly those jobs, of every workflow type, and writes a receipt of what stopped and what could not be proven stopped, which management signs with the audit key. The command prints the receipt's ID (oz0 receipt show).
Setting a switch that holds already returns it; with --mode cancel on a parked one, its jobs are cancelled instead. A switch a cluster admin set only a cluster admin may release or change. The reason is at most 500 characters.
Usage
oz0 kill [command]
Subcommands
Flags
| Flag | Type | Default | Description |
|---|---|---|---|
--tenant | string | default | the tenant |
Inherited flags
| Flag | Type | Default | Description |
|---|---|---|---|
--context | string | $OZ0_CONTEXT or <config dir>/orchestrator-zero/context.json | CLI context file written by server init |
--log-level | string | info | log level: debug, info, warn or error |
See all oz0 commands and the global flags.