Exit Status

Scripts branch on sind’s exit status, so it is part of the command line contract: a status may be added, but never renumbered.

Status	Meaning	Typical cause
`0`	The command succeeded
`1`	The command ran and failed	Docker is not reachable, a node did not become ready within `--wait`, a cluster does not exist, a required `sind doctor` check failed, an invalid `SIND_REALM` or config file
`2`	sind rejected the command line before acting on it	An unknown command or flag, a flag value or argument that is not valid, the wrong number of arguments
`130`	SIGINT or SIGTERM interrupted the command	Ctrl-C, `timeout`, `docker stop`

sind ssh, sind exec, sind enter and sind logs exit with the status of the program they run instead, see below.

Usage errors

Status 2 means that the command line has to change before the command can succeed; running it again will not help.

$ sind get cluster dev extra
12:00:00.000 ERRO accepts at most 1 arg(s), received 2
$ echo $?
2

This covers:

an unknown command or help topic: sind get bogus, sind help bogus, sind completion fishy
an unknown flag, a flag without its value, or a value that is not valid: sind get clusters -o yaml, sind --realm Not_A_Realm get clusters
the wrong number of arguments, or an argument that is not valid: sind get cluster a b, sind get cluster Not_A_Name
a node argument that does not expand, or that names several nodes where one is needed: sind power on 'worker-[', sind ssh 'worker-[0-1]'
sind exec without -- and a command, and sind ssh with a remote command that does not follow --: sind ssh worker-0 hostname

SIND_REALM and the config file are not part of the command line: an invalid one exits 1.

Commands that run a program

sind ssh, sind exec, sind enter and sind logs hand the terminal to a program, docker exec running ssh, the command or a shell, or docker logs, and exit with its status. They add no error line of their own, since the program has written its own diagnostics.

$ sind exec -- sh -c 'exit 42'
$ echo $?
42

The status is the command’s, as docker exec reports it, and for sind ssh the remote command’s, as ssh reports it. ssh exits 255 when it cannot connect. Docker’s own errors, such as a container that is not running, give docker’s status and message.
A failure before the program starts is sind’s: sind exec in a cluster without a submitter or controller exits 1, and a usage error exits 2.
A docker process killed by signal N exits 128 + N, as a shell reports it. Killed by SIGINT or SIGTERM, it exits 130, like an interrupted sind: these signals reach docker when they are sent to sind’s process group, as by Ctrl-C.

Interrupts

The first SIGINT or SIGTERM stops the command and runs its cleanup, such as the rollback of a failed sind create cluster; sind then exits 130. A --wait limit of sind create cluster or sind create worker that runs out is not an interrupt: the command rolls back the same way and exits 1. SIGTERM exits 130 too, not 143, so that a script checks one status for “interrupted”. A second signal ends sind at once, without waiting for the cleanup to finish.

MCP tools

A tool call of the MCP server returns the command’s exit status as exitCode, next to its stdout and stderr: a tool call whose arguments sind rejects reads 2, and sind_exec returns the status of the command it ran.