Monitoring TAS App Lifecycle Metrics vs. Cloud Controller Audit Events

Products

• VMware Tanzu Platform Core

Issue/Introduction

In Tanzu Application Service (TAS), application lifecycle actions (Start, Stop, Restart, Update, Deployment) and process crashes are captured differently across the platform telemetry layer. Only container crashes generate native runtime metrics, while lifecycle updates are recorded as database state events.

Lifecycle Event	Identified Name	Signal Type	Origin Subsystem	Storage / Location	OTel Native Metric?	Tanzu Hub
Crash	crashed_actual_lrps / app.crash	Metric / Event	Diego BBS (diego_database)	Loggregator / Firehose	Yes	App Metrics & Events
Startup	audit.app.start	State Event	Cloud Controller	CC DB (app_usage_events)	No	App Events Table
Shutdown	audit.app.stop	State Event	Cloud Controller	CC DB (app_usage_events)	No	App Events Table
Restart	audit.app.restart	State Event	Cloud Controller	CC DB (app_usage_events)	No	App Events Table
Update	audit.app.update	State Event	Cloud Controller	CC DB (app_usage_events)	No	App Events Table
Deployment	audit.app.deployment.creates	State Event	Cloud Controller	CC DB (app_usage_events)	No	App Events Table

 

Environment

Tanzu Application Service (TAS)

Resolution

1. Diego BBS Metric Monitoring
   crashed_actual_lrps is the only item in this list emitted natively as a performance metric. It is monitored and exported by the Diego Bulletin Board Server (BBS) within the diego_database subsystem to track physical, failed Long Running Process (LRP) container instances.

2. Cloud Controller Audit Records
   Lifecycle actions (audit.app.start, audit.app.stop, audit.app.restart, audit.app.update, and audit.app.deployment.creates) represent state-changing administrative operations. They are recorded directly by the Cloud Controller into its internal relational database under the app_usage_events table, rather than being emitted as real-time metric counters.

3. OpenTelemetry and Tanzu Hub Behavior
   OpenTelemetry System: Because Cloud Controller audit records are stored as database usage events rather than timeseries metrics, they are not emitted as native metric streams through standard OpenTelemetry collectors.
   Tanzu Hub: Tanzu Hub queries these events directly from Cloud Controller records and displays them in the App Events table, rather than populating them from OpenTelemetry metric pipelines.