Skip to content

Container Apps Diagnostics (Revision and Startup Analysis)

Analyze Azure Container Apps platform and console logs to identify unhealthy revisions, startup failures, and scaling activity that explains why a deployment is not serving traffic as expected.

Scenario

You need to identify revisions with repeated startup or scale events in the last 24 hours and determine whether the issue is caused by the app container or the managed environment.

KQL Query

ContainerAppSystemLogs_CL
| where TimeGenerated > ago(24h)
| summarize
    EventCount = count(),
    LastSeen = max(TimeGenerated),
    arg_max(TimeGenerated, Log_s)
    by ContainerAppName_s, RevisionName_s, Reason_s
| where Reason_s in ("RevisionProvisioningError", "ContainerCreateFailure", "HealthCheckFailed", "ScaleRuleTriggered", "DaprSidecarUnhealthy")
| project
    ContainerApp = ContainerAppName_s,
    Revision = RevisionName_s,
    Reason = Reason_s,
    EventCount,
    LastSeen,
    LatestMessage = Log_s
| order by EventCount desc, LastSeen desc
| take 15

Data Flow

graph TD
    A[ContainerAppSystemLogs_CL table] --> B[Filter by 24h]
    B --> C[Group by app, revision, and reason]
    C --> D[Keep startup and scaling failures]
    D --> E[Project latest message context]
    E --> F[Review unhealthy revisions first]

Sample Output

ContainerApp Revision Reason EventCount LastSeen LatestMessage
aca-orders aca-orders--green HealthCheckFailed 11 2026-04-13 09:36:00Z Readiness probe failed on port 8080 after revision activation
aca-orders aca-orders--green ScaleRuleTriggered 7 2026-04-13 09:35:00Z HTTP scale rule increased replica count from 2 to 5

How to Read This

HealthCheckFailed or ContainerCreateFailure usually indicates the revision never became healthy enough to receive stable traffic. ScaleRuleTriggered by itself is not a failure, but when it appears alongside startup errors it often means traffic is increasing while new replicas fail to initialize fast enough.

Limitations

  • Azure Container Apps diagnostic logs must be routed to Log Analytics before these tables appear.
  • Table names and columns can vary slightly by ingestion mode or workspace schema evolution.
  • Console-level application errors often require a companion query against ContainerAppConsoleLogs_CL for full stack traces.

See Also

Sources