This runbook provides operational procedures for the InventoryService, which is responsible for managing product inventory and stock levels across the FlowMart e-commerce platform.

Architecture

The InventoryService is responsible for:

• Managing product inventory and stock levels
• Reserving inventory for pending orders
• Tracking inventory across warehouses and locations
• Providing real-time availability information
• Triggering restock notifications

Service Dependencies

Monitoring and Alerting

Key Metrics

Dashboards

• InventoryService Overview
• Stock Level Alerts
• Database Performance

Common Alerts

Troubleshooting Guides

High Latency

If the service is experiencing high latency:

1. Check system resource usage:

   kubectl top pods -n inventory

2. Check database connection pool:

   kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl localhost:8080/actuator/metrics/hikaricp.connections.usage

3. Check cache hit rate:

   kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl localhost:8080/actuator/metrics/cache.gets | grep "hit_ratio"

4. Check for slow queries in the database:

   kubectl exec -it $(kubectl get pods -l app=postgresql -n data -o jsonpath='{.items[0].metadata.name}') -n data -- psql -U postgres -c "SELECT query, calls, mean_exec_time FROM pg_stat_statements ORDER BY mean_exec_time DESC LIMIT 10;"

5. Scale the service if needed:

   kubectl scale deployment inventory-service -n inventory --replicas=5

Database Issues

If there are database connection or performance issues:

1. Check PostgreSQL status:

   kubectl exec -it $(kubectl get pods -l app=postgresql -n data -o jsonpath='{.items[0].metadata.name}') -n data -- pg_isready -U postgres

2. Check for long-running transactions:

   kubectl exec -it $(kubectl get pods -l app=postgresql -n data -o jsonpath='{.items[0].metadata.name}') -n data -- psql -U postgres -c "SELECT pid, now() - xact_start AS duration, state, query FROM pg_stat_activity WHERE state != 'idle' ORDER BY duration DESC;"

3. Check for table bloat:

   kubectl exec -it $(kubectl get pods -l app=postgresql -n data -o jsonpath='{.items[0].metadata.name}') -n data -- psql -U postgres -c "SELECT schemaname, relname, n_live_tup, n_dead_tup, (n_dead_tup::float / n_live_tup::float) AS dead_ratio FROM pg_stat_user_tables WHERE n_live_tup > 1000 ORDER BY dead_ratio DESC;"

4. Restart database connections in the application if needed:

   kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl -X POST localhost:8080/actuator/restart-db-connections

Cache Issues

If there are Redis cache issues:

1. Check Redis status:

   kubectl exec -it $(kubectl get pods -l app=redis -n data -o jsonpath='{.items[0].metadata.name}') -n data -- redis-cli ping

2. Check Redis memory usage:

   kubectl exec -it $(kubectl get pods -l app=redis -n data -o jsonpath='{.items[0].metadata.name}') -n data -- redis-cli info memory

3. Check cache hit rate:

   kubectl exec -it $(kubectl get pods -l app=redis -n data -o jsonpath='{.items[0].metadata.name}') -n data -- redis-cli info stats | grep hit_rate

4. Clear cache if necessary:

   kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl -X POST localhost:8080/actuator/caches/clearAll

Stock Management

For critical stock issues:

1. Identify products with low or no stock:

   kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl localhost:8080/internal/api/inventory/low-stock

2. Check for stuck inventory reservations:

   kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl localhost:8080/internal/api/inventory/stuck-reservations

3. Release expired reservations if necessary:

   kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl -X POST localhost:8080/internal/api/inventory/release-expired-reservations

4. Manually update inventory levels for emergency corrections:

   curl -X PUT https://api.internal.flowmart.com/inventory/products/{productId}/stock \
     -H "Authorization: Bearer $ADMIN_TOKEN" \
     -H "Content-Type: application/json" \
     -d '{"warehouseId": "WAREHOUSE_ID", "quantity": 100, "reason": "Manual correction"}'

Common Operational Tasks

Scaling the Service

To scale the service horizontally:

kubectl scale deployment inventory-service -n inventory --replicas=<number>

Restarting the Service

To restart all pods:

kubectl rollout restart deployment inventory-service -n inventory

Database Maintenance

For routine database maintenance:

1. Run VACUUM ANALYZE to optimize tables:

   kubectl exec -it $(kubectl get pods -l app=postgresql -n data -o jsonpath='{.items[0].metadata.name}') -n data -- psql -U postgres -c "VACUUM ANALYZE inventory_items;"

2. Update database statistics:

   kubectl exec -it $(kubectl get pods -l app=postgresql -n data -o jsonpath='{.items[0].metadata.name}') -n data -- psql -U postgres -c "ANALYZE;"

Reconcile Inventory

To reconcile inventory with the warehouse management system:

kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl -X POST localhost:8080/internal/api/inventory/reconcile

Manually Trigger Restock Notifications

To trigger restock notifications for low stock items:

kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl -X POST localhost:8080/internal/api/inventory/trigger-restock-notifications

Recovery Procedures

Database Failure Recovery

If the PostgreSQL database becomes unavailable:

1. Verify the status of the PostgreSQL cluster:

   kubectl get pods -l app=postgresql -n data

2. If the primary instance is down, check if automatic failover has occurred:

   kubectl exec -it $(kubectl get pods -l app=postgresql-patroni -n data -o jsonpath='{.items[0].metadata.name}') -n data -- patronictl list

3. If automatic failover has not occurred, initiate manual failover:

   kubectl exec -it $(kubectl get pods -l app=postgresql-patroni -n data -o jsonpath='{.items[0].metadata.name}') -n data -- patronictl failover

4. Once database availability is restored, validate the InventoryService functionality:

   curl -X GET https://api.internal.flowmart.com/inventory/health

Cache Failure Recovery

If the Redis cache becomes unavailable:

1. Verify Redis cluster status:

   kubectl get pods -l app=redis -n data

2. If needed, restart the Redis cluster:

   kubectl rollout restart statefulset redis -n data

3. The InventoryService will fall back to database queries when the cache is unavailable.

4. When the cache is restored, you can warm it up:

   kubectl exec -it $(kubectl get pods -l app=inventory-service -n inventory -o jsonpath='{.items[0].metadata.name}') -n inventory -- curl -X POST localhost:8080/internal/api/inventory/warm-cache

Disaster Recovery

Complete Service Failure

In case of a complete service failure:

1. Initiate incident response by notifying the on-call team through PagerDuty.

2. Verify the deployment status:

   kubectl describe deployment inventory-service -n inventory

3. If necessary, restore from a previous version:

   kubectl rollout undo deployment inventory-service -n inventory

4. If the primary region is experiencing issues, fail over to the secondary region:

   ./scripts/dr-failover.sh inventory-service

5. Verify the service is functioning in the secondary region:

   curl -X GET https://api-dr.internal.flowmart.com/inventory/health

Maintenance Tasks

Deploying New Versions

kubectl set image deployment/inventory-service -n inventory inventory-service=ecr.aws/flowmart/inventory-service:$VERSION

Database Schema Updates

For database schema updates:

1. Notify stakeholders through the #maintenance Slack channel.

2. Set InventoryService to maintenance mode:

   curl -X POST https://api.internal.flowmart.com/inventory/admin/maintenance -H "Authorization: Bearer $ADMIN_TOKEN" -H "Content-Type: application/json" -d '{"maintenanceMode": true, "message": "Database schema update"}'

3. Apply the database migrations:

   kubectl apply -f inventory-flyway-job.yaml

4. Verify migration completion:

   kubectl logs -l job-name=inventory-flyway-migration -n inventory

5. Turn off maintenance mode:

   curl -X POST https://api.internal.flowmart.com/inventory/admin/maintenance -H "Authorization: Bearer $ADMIN_TOKEN" -H "Content-Type: application/json" -d '{"maintenanceMode": false}'

Contact Information

Primary On-Call: Inventory Team (rotating schedule)
Secondary On-Call: Platform Team
Escalation Path: Inventory Team Lead > Engineering Manager > CTO

Slack Channels:

• #inventory-support (primary support channel)
• #inventory-alerts (automated alerts)
• #incident-response (for major incidents)

Reference Information

• InventoryService API Documentation
• Architecture Diagram
• Service Level Objectives (SLOs)
• Database Schema