Skip to content

Replica Count Over Time¶

Use this query to visualize how many replicas are active per revision over time and detect scaling plateaus.

Data Source¶

Table	Schema Note
`ContainerAppSystemLogs_CL`	Legacy schema. If empty, try `ContainerAppSystemLogs` (non-`_CL`).

Query Pipeline¶

flowchart LR
    A[Filter by app and time] --> B[Keep replica lifecycle events] --> C[Distinct replica count by bin] --> D[Trend output]

Query¶

let AppName = "my-container-app";
let Window = 12h;
ContainerAppSystemLogs_CL
| where ContainerAppName_s == AppName and TimeGenerated >= ago(Window)
| where isnotempty(ReplicaName_s)
| summarize replicas=dcount(ReplicaName_s) by bin(TimeGenerated, 5m), RevisionName_s
| order by TimeGenerated asc

Example Output¶

TimeGenerated	RevisionName_s	replicas
2026-04-04T11:30:00.000Z	ca-myapp--0000001	1
2026-04-04T11:35:00.000Z	ca-myapp--0000001	2
2026-04-04T11:40:00.000Z	ca-myapp--0000001	3
2026-04-04T11:45:00.000Z	ca-myapp--0000001	2

Interpretation Notes¶

Flat low replicas under high load suggests scaling bottleneck.
Sudden drops may indicate crash loops or rollbacks.
Normal pattern: smooth increase/decrease aligned with traffic profile.

Limitations¶

Distinct count from logs is approximate and ingestion-dependent.
Best used with platform metrics for exact replica counts.

See Also¶