azure-observability-baseline

# azure-observability-baseline **Impact tier**: P1 -- Operational Readiness ## Purpose Close the observability gap where the Odoo runtime lacks Azure-native monitoring. The benchmark audit found: no Application Insights instrumentation, no Log Analytics workspace, no Azure Monitor alert rules, and residual Docker/Prometheus assumptions in documentation that do not apply to the ACA runtime. ## When to Use - Instrumenting Odoo on Azure Container Apps for the first time. - Replacing Docker-centric monitoring (Prometheus, cAdvisor) with Azure-native stack. - Setting up alert rules for Odoo health, PG health, and ACA scaling events. - Preparing for go-live observability readiness. ## Required Evidence (inspect these repo paths first) | Path | What to look for | |------|-----------------| | `infra/azure/modules/app-insights.bicep` | App Insights resource, connection string output | | `infra/azure/odoo-runtime.bicep` | ACA env diagnostic settings, App Insights binding | | `infra/azure/main.bicep` | Log Analytics workspace, module wiring | | `infra/ssot/azure/resources.yaml` | Log Analytics, App Insights entries | | `docs/runbooks/ODOO18_GO_LIVE_CHECKLIST.md` | Monitoring/alerting line items | | `docs/audits/ODOO_AZURE_ENTERPRISE_BENCHMARK.md` | Observability gap row | ## Microsoft Learn MCP Usage Run at least these three queries: 1. `microsoft_docs_search("Azure Container Apps Application Insights monitoring")` -- retrieves how to bind App Insights to ACA environment. 2. `microsoft_docs_search("Azure Monitor alert rules container apps CPU memory")` -- retrieves alert rule configuration for ACA metrics. 3. `microsoft_docs_search("Azure Log Analytics workspace diagnostic settings PostgreSQL")` -- retrieves how to route PG Flexible Server logs to Log Analytics. Optional: 4. `microsoft_code_sample_search("bicep application insights log analytics workspace", language="bicep")` 5. `microsoft_docs_fetch("https://learn.microsoft.com/en-us/azure/container-apps/observability")` ## Workflow 1. **Inspect repo** -- Read Bicep modules for existing App Insights and Log Analytics resources. Check if ACA environment has `appInsightsConfiguration`. Search for any Prometheus/Docker monitoring references to remove. 2. **Query MCP** -- Run the three searches. Capture App Insights connection string binding for ACA, alert rule metric names, PG diagnostic categories. 3. **Compare** -- Identify: (a) Does a Log Analytics workspace exist in Bicep? (b) Is App Insights connected to ACA? (c) Are alert rules defined? (d) Are there stale Prometheus references? 4. **Patch** -- Update Bicep: - Create Log Analytics workspace if missing. - Create App Insights connected to that workspace. - Bind App Insights to ACA environment via `appInsightsConfiguration`. - Add diagnostic settings for PG Flexible Server. - Add alert rules: ACA CPU > 80%, ACA memory > 80%, PG connections > 80%, Odoo health check failure. - Remove Docker/Prometheus monitoring references from docs. 5. **Verify** -- Bicep lints clean. SSOT updated. Go-live checklist has monitoring items. No Prometheus/Docker monitoring references remain. ## Outputs | File | Change | |------|--------| | `infra/azure/modules/app-insights.bicep` | App Insights + Log Analytics | | `infra/azure/odoo-runtime.bicep` | ACA App Insights binding | | `infra/azure/modules/postgres-flexible.bicep` | Diagnostic settings | | `infra/azure/modules/alerts.bicep` | Alert rules (new file) | | `infra/ssot/azure/resources.yaml` | Log Analytics, App Insights entries | | `docs/runbooks/ODOO18_GO_LIVE_CHECKLIST.md` | Monitoring verification steps | | `docs/evidence/<stamp>/azure-observability-baseline/` | Bicep diffs, MCP excerpts | ## Completion Criteria - [ ] Log Analytics workspace exists in Bicep with 30-day retention minimum. - [ ] Application Insights is created and connected to Log Analytics. - [ ] ACA environment has `appInsightsConfiguration.connectionString` set. - [ ] PG Flexible Server has diagnostic settings routing to Log Analytics. - [ ] At least 4 alert rules defined (CPU, memory, PG connections, health check). - [ ] No Prometheus, Grafana, or cAdvisor references remain in `docs/runbooks/`. - [ ] Evidence directory contains Bicep diffs and MCP query excerpts.