abc doctor

Verifies that the CLI is correctly configured and can reach and submit work to the active cluster. Where abc cluster status (grove+ tiers) asks the server "are your services healthy?", abc doctor starts from your ~/.abc/config.yaml and proves the full client→alloc path end to end — the check you run first when something isn't working.

Usage

abc doctor                      # all three check groups
abc doctor --skip-job           # config + connectivity only (no probe job)
abc doctor --context=seedling   # check a named context, not the active one
abc doctor --json               # machine-readable result object
abc doctor --job-timeout=2m     # allow longer for the probe job to complete

Check groups

Three groups run in order; a failure in an earlier group still lets later groups run (so you see the full picture, not just the first error).

#	Group	What it checks
1	Config	config.yaml loads, an active context is set, the Nomad address and access token are present.
2	Connectivity	Nomad is reachable and reports healthy. MinIO/S3 is probed when the active context has an S3 endpoint configured.
3	Workload	Submits a minimal raw_exec batch probe job, waits up to the job timeout (default 60 s) for it to complete, then purges it. Proves submission + scheduling + completion actually work, not just connectivity. Skip with --skip-job.

Flags

Flag	Default	Description
--skip-job	false	Run groups 1 and 2 only — skip the workload probe. Useful on a constrained cluster or when you only want to validate config + reachability.
--context=<name>	active context	Run the checks against a named context instead of the active one.
--json	false	Emit results as a single JSON object (one entry per check with pass/fail + detail).
--job-timeout=<dur>	1m	Maximum time to wait for the probe job to complete before declaring the workload check failed.

Exit codes

Code	Meaning
0	All run checks passed.
1	One or more checks failed.

abc doctor vs abc cluster status

	abc doctor	abc cluster status
Starting point	your local config.yaml	the cluster's service registry
Proves	the full client → Nomad → alloc submission path	server-side service health
Submits a job?	yes (the workload probe) unless --skip-job	no
Use when	"my CLI can't talk to the cluster — why?"	"is the cluster itself healthy?"

See also

• abc cluster status (grove+ tiers) — server-side service health table
• Global flags — --context, --json, --quiet