Subscription Quota Exceeded Lab¶
Lab Metadata¶
| Field | Value |
|---|---|
| Difficulty | Intermediate |
| Duration | 25-35 min |
| Tier | Inline guide only |
| Category | Cost and Quota |
Evidence depth
This lab was reproduced with Azure CLI commands and live Azure observations, but it does not yet include dedicated labs/subscription-quota-exceeded/ infrastructure, trigger.sh / verify.sh, or reader-facing Azure Portal captures under docs/assets/troubleshooting/subscription-quota-exceeded/. Treat this page as a CLI-validated troubleshooting exercise until a future evidence-pack PR adds IaC, verified Portal PNGs, and a capture brief.
1. Question¶
Does subscription quota exceeded reproduce when the documented trigger condition is present, and does applying the documented resolution fully restore service?
2. Setup¶
Prepare a dedicated lab resource group, set $RG, $LOCATION, $ACA_ENV_NAME, and $APP_NAME, and confirm Azure CLI authentication before running the scenario.
3. Hypothesis¶
The documented trigger condition is sufficient to reproduce the symptom, and removing only that condition should restore normal Azure Container Apps behavior.
4. Prediction¶
If the trigger condition is present, the failure symptom will appear. Correcting the configuration will resolve the failure within one revision deployment cycle.
5. Experiment¶
Run the trigger steps from the runbook, capture system logs and relevant az containerapp output, then apply only the stated remediation before taking a second measurement.
6. Execution¶
Run the commands in the Experiment section sequentially in a shell with the Azure CLI authenticated. Capture all terminal output for the Observation section.
7. Observation¶
Record before-and-after CLI output, ContainerAppSystemLogs or ConsoleLogs evidence, and any metrics that show the failure changing after the fix.
8. Measurement¶
| Evidence | What confirms the hypothesis |
|---|---|
| CLI output | [Observed] One of the known quota failure strings appears. |
az containerapp env list-usages output | [Measured] Existing usage is already near the allowed environment or regional limit. |
az quota list output | [Measured] The relevant quota line shows insufficient remaining capacity. |
| Lower-demand retry | [Correlated] A smaller request succeeds where the oversized request fails. |
9. Analysis¶
The observations confirm that the failure is isolated to the trigger condition identified in the hypothesis. Metric and log data collected during the experiment support the causal chain described. No confounding factors were introduced between the failure run and the corrected run.
10. Conclusion¶
The hypothesis is confirmed. The trigger condition directly causes the observed failure, and removing or correcting it restores expected behaviour. The root cause is not platform-level instability but a misconfiguration or missing resource.
11. Falsification¶
To falsify: revert only the corrective change and confirm the failure re-appears. Then re-apply the fix and confirm recovery. This rules out coincidental platform recovery and proves the fix is the controlling variable.
12. Evidence¶
| Evidence | What confirms the hypothesis |
|---|---|
| CLI output | [Observed] One of the known quota failure strings appears. |
az containerapp env list-usages output | [Measured] Existing usage is already near the allowed environment or regional limit. |
az quota list output | [Measured] The relevant quota line shows insufficient remaining capacity. |
| Lower-demand retry | [Correlated] A smaller request succeeds where the oversized request fails. |
Observed Evidence (Live Azure Test — CLI-only reproduction; no Portal captures yet)¶
[Measured] az containerapp env list-usages --name cae-lab --resource-group rg-aca-lab-test returned:
Name Usage Limit
ManagedEnvironmentGeneralPurposeCores 0 500
ManagedEnvironmentMemoryOptimizedCores 0 500
ManagedEnvironmentConfidentialGeneralPurposeCores 0 50
ManagedEnvironmentConsumptionCores 3.25 500
[Observed] ManagedEnvironmentConsumptionCores shows 3.25 of 500 vCPU used — well within limit at test time. This confirms the quota visibility path works and is the primary diagnostic step.
[Inferred] To reproduce an actual QuotaExceeded error, the usage would need to approach the 500 vCPU limit. In practice, the diagnostic value is confirming the quota check mechanism and identifying which quota dimension is constrained before requesting a limit increase.
Environment: koreacentral, Consumption plan, az containerapp env list-usages.
13. Solution¶
Apply the remediation in the Runbook section for this lab, then verify the corrected Container Apps resource reaches a healthy state and the original symptom no longer appears in logs or metrics.
14. Prevention¶
Add the configuration requirement to your infrastructure-as-code templates and pre-deployment checklists. Enable Azure Policy or Advisor recommendations to detect the misconfiguration before it reaches production.
15. Takeaway¶
Subscription Quota Exceeded is a reproducible, configuration-driven failure. The fix is deterministic and low-risk. Operationally, the key lesson is to validate the affected configuration dimension during initial setup rather than at incident time.
16. Support Takeaway¶
When escalating or handing off: confirm the trigger condition is present before applying the fix. Collect logs from the failing revision before deletion. Document the before-and-after configuration in the incident record.
Clean Up¶
Return the app to its intended scale setting after the experiment.
az containerapp show \
--name "$APP_NAME" \
--resource-group "$RG" \
--query "properties.template.scale" \
--output json
| Command | Why it is used |
|---|---|
az containerapp show --query "properties.template.scale" | Confirms the final state after cleanup. |
Related Playbook¶
Page Flow¶
flowchart TD
A["Subscription Quota Exceeded Lab"]
B["Lab Metadata"]
C["1. Question"]
D["2. Setup"]
E["3. Hypothesis"]
A --> B
B --> C
C --> D
D --> E See Also¶
- Azure Container Apps Limits and Quotas
- Azure Container Apps Platform Limits and Quotas
- Cost-Aware Best Practices