Service Monitoring #

Platform Operators monitor the service fleet through Service Topology and Operations in https://admin.<your-domain>, backed by control-backendd and host-ops-agent.

Allowlisted Services (host-ops-services.json) #

Only services explicitly listed in TSUBAME_HOST_OPS_MANIFEST_PATH are visible or controllable:

Service IDDisplay NameRestartableHealth Endpoint
postgresPostgreSQLYes (high-risk)pg_isready
identitydIdentityYes (high-risk)/proc/1/cmdline / gRPC :50051
tenant-configdTenant ConfigurationYesgRPC :50052 / HTTP :8082
control-backenddControl BackendYes (high-risk)GET /healthz, GET /readyz
scannerdAttachment ScannerYesInternal scanner_isolated_net
mail-workerMail WorkerYesWorker loop health
smtp-inSMTP InboundYes (high-risk)GET /healthz, GET /readyz
imapdIMAP ServerYesGET /healthz
pop3dPOP3 ServerYesGET :8087/healthz
mta-outOutbound MTAYes (high-risk)GET :8088/healthz
host-ops-docker-proxyHost Operations ProxyNoInternal bridge
host-ops-agentHost Operations AgentNohttp://host-ops-agent:8082

Monitoring & Restart Endpoints #

  • List Services: GET /v1/platform/services (requires platform.ops.read).
  • Restart Service: POST /v1/platform/services/:id/restart (requires platform.ops.restart, idempotency key, non-empty reason, and Step-Up confirmation for high-risk core services).
  • Track Restart Job: GET /v1/platform/operations/restarts/:jobId.
WARNING

host-ops-docker-proxy and host-ops-agent are marked restartable: false in the manifest and cannot be restarted remotely via the API.