Status reporter

StatusReporter is the in-memory aggregation hub for module health and telemetry. It is started first among Supervisor modules and backs GET /v1/system/status, Field Agent status POST payloads, and portions of system info.

Code: internal/statusreporter/

Purpose

• Hold structured status objects for each major module
• Track daemon lifecycle (STARTING, RUNNING, WARNING, ...)
• Maintain modulesStatus[] array (fixed indexes; Resource Manager slot unused) for Supervisor reporting
• Background worker for system time synchronization
• Assemble composite status for API and Controller export

Dependencies

Depends on	Reason
config	Hostname, version metadata
store	Optional runtime class listing for status enrichment

Used by	Reason
All major modules	Update*Status() callbacks
edgeletapi/handlers/status.go	System status HTTP response
fieldagent	Status POST body construction
resourceconsumption	Writes CPU/memory/disk usage

Lifecycle

Start

First module in Supervisor start order:

1. Initialize default status structs (NewSupervisorStatus(7), etc.)
2. Start setSystemTimeWorker() goroutine

Stop

Cancel context; wait for worker.

Module index mapping

Supervisor updates SupervisorStatus.modulesStatus[i] using constants from internal/utils/constants.go:

Index	Module
0	Resource Consumption Manager
1	Process Manager
2	Status Reporter
3	EdgeletAPI
4	Field Agent
5	(unused: former Resource Manager; indexes are not reshuffled)
6	GPS Manager

SupervisorStatus fields: daemonStatus, daemonLastStart, operationDuration, warningMessage.

Status object types

Getter	Model type
GetSupervisorStatus	SupervisorStatus
GetProcessManagerStatus	ProcessManagerStatus
GetFieldAgentStatus	FieldAgentStatus
GetEdgeletAPIStatus	EdgeletAPIStatus
GetResourceConsumptionManagerStatus	Resource usage
GetSSHProxyManagerStatus	SSH tunnel state
GetVolumeMountManagerStatus	Active mount count

Process Manager status includes the per-microservice map (runtime state, errorMessage, lastError, lastErrorAt, restartCount), registry status, and running count.

Current errorMessage stays set until 30 seconds of continuous RUNNING, then is sent as "". lastError is the last crash text and is not cleared on recovery. For controller-managed workloads those last-crash fields are in-memory (lost on agent restart until the next failure). Local inspect may also surface SQLite last_error / restart_count.

External APIs

Surface	Role
GET /v1/system/status	Primary operator view
Controller status POST	Field Agent aggregates reporter state

External reporter helpers in reporter_external.go format Controller-compatible payloads.

Configuration

No direct config keys; updated via module callbacks on config reload.

Observability

• Log module: "Status Reporter"
• Module index: 2 (Status Reporter itself in the array)

Failure modes

Symptom	Typical cause
Stale module status	Module failed to update on stop
WARNING stuck	Edge Guard or engine degraded; check warningMessage
Missing MS in status	Process Manager not publishing map
Last crash gone after edgelet restart	Controller-managed last-error is in-memory; next inspect/exit fills it
Dashboard still shows a crash after recovery	Grace window: errorMessage clears only after 30s continuous RUNNING

Code map

File	Role
reporter.go	Singleton, getters/updaters, system time worker
reporter_external.go	Controller export formatting

Related: Supervisor, Field agent, Edgelet API module.

See anything wrong with the document? Help us improve it!

Edit this page on GitHub!