Penpot Backend Connection Issues – Database Rollback Fail

Self-host Back End
1

I’m running self-hosted Docker Penpot and am seeing backend errors related to database transactions failing due to connection issues.

Details:

  • Penpot Backend: Running in a Docker Compose stack on one machine.
  • Database: PostgreSQL is on a separate VM in my VM network but not part of the Docker Compose stack.
  • Network: The database and Penpot communicate over a local VM network.

Error Summary:

  • The backend worker is failing with an I/O error while sending to the backend, leading to rollback failures in transactions.
  • Logs show that the connection is reset, causing transactions to fail and preventing rollbacks from succeeding.
  • The issue originates from next.jdbc.transaction while interacting with PostgreSQL.
  • HikariCP reports that the connection is closed, preventing rollback from completing.
  • The PostgreSQL logs show “An I/O error occurred while sending to the backend”.

Questions:

  • Has anyone encountered Penpot losing database connections like this?
  • Could this be a timeout issue, connection pooling misconfiguration problem?
  • Any recommendations for debugging PostgreSQL and HikariCP connection handling in a separate database VM setup?

Any insights would be greatly appreciated! Thanks!