Diagnostic workflows

The command reference is organised by what each command does. These are organised by what has gone wrong.

Symptom	Workflow
The site never came online, or dropped out of the tenant	Registration
Routes are not advertised or learned; the VIP is unreachable	BGP
Traffic reaches the node but does not arrive	Data plane

Each step links to the page that documents the command. Nothing is restated here, so the syntax has exactly one home and cannot drift out of agreement with the node.

Before any of them

Establish which access path you are on, because it changes what is possible:

curl -sS -H "Authorization: APIToken $XCSH_API_TOKEN" \
  "$XCSH_API_URL/api/config/namespaces/system/sites/<site>" | jq -r '.spec.site_state'

ONLINE means the debug API will answer. Anything else means the registration workflow is the one you want, and one of the routes that does not depend on registration: Site Console while the node still has a network path, serial console when it does not.

A site that is missing entirely has one more explanation to rule out before “it never registered”: that you are querying the wrong tenant. cd terraform && terraform output -raw xc_tenant names the one this deployment belongs to.