[PostgreSQL] Intermittent i/o timeout errors and request for better connection pool configuration & metrics #1107
Description
Hello!
We are experiencing intermittent i/o timeout errors on established connections to our external PostgreSQL instance. These failures affect core operations: DELETE, INSERT (add), and UPDATE, impacting the reliability of policy-reporter.
Error Logs
The logs show i/o timeout errors on an already established TCP connection during different operations:
1.7556721861647866e+09 error failed to remove previous policy report {"error": "read tcp 10.XXX.XXX.XXX:43690->10.XXX.XXX.XXX:5432: i/o timeout"}
1.7556721861648421e+09 error failed to add policy report {"name": "polr-ns-test", "error": "read tcp 10.XXX.XXX.XXX:43690->10.XXX.XXX.XXX:5432: i/o timeout"}
1.7557270106546857e+09 error failed to update policy report {"name": "110d8ed3-1159-11e8-9d6b-005056a934e4", "error": "read tcp 10.XXX.XXX.XXX:43690->10.XXX.XXX.XXX:5432: i/o timeout"}
We see that the application uses the Bun ORM, which has a built-in connection pool. However, it's unclear how to configure its parameters (like timeouts, pool size) through the Helm chart. Furthermore, we lack visibility into the database layer performance and error rates.
Environment Details:
Policy-Reporter Version: 3.1.3
Database: External PostgreSQL
Deployment Environment: Kubernetes (via Helm chart - 3.1.3)
We would like to request:
- Exposure of Bun's connection pool settings via Helm chart values.
- Integration of Bun's monitoring hooks to expose database-related metrics.