Docker: Hydra connection pool exhaustion when history is disabled

[k1bac]

Environment details:

• NZBHydra2 version: 8.3.0
• Docker image: linuxserver/nzbhydra2:v8.3.0-ls74 (Release, by linuxserver.io)
• Host OS: Debian Trixie

---

Expected and actual behavior:

• Expected: Hydra should remain operational indefinitely when history is disabled. Cleanup tasks should not exhaust the database connection pool.
• Actual: After some uptime (days to a week), all database-related actions fail with connection pool exhaustion errors. Restarting the container resolves the issue temporarily.

---

Reproducibility:

• Problem occurs consistently if history is disabled ("Keep history" off).
• Does not appear if history is enabled, even with a short retention period (e.g., 1 week).
• Moderate usage, reproducible over time.

---

Steps to reproduce:

1. Run nzbhydra2 in Docker (linuxserver/nzbhydra2:v8.3.0-ls74)
2. In settings, disable history (Keep history: off)
3. Use as normal for several days under moderate/typical load
4. Eventually, these errors occur:
   • Could not open JPA EntityManager for transaction (Error Code: 900)
   • Unable to acquire JDBC Connection [HikariPool-1 - Connection is not available, request timed out after 30000ms.]
   • Could not open JPA EntityManager
5. Restarting the container always clears the error for a while.

---

GUI info:
Not a GUI bug

---

Searched for problem:

• Yes, checked previous issues (including Hydra2 timeout #289); this is specific to Docker with history disabled.

---

Let me know if you need further info or logging at debug/trace level. Can keep container running in broken state for diagnostics if required.

This issue was written with Copilot, findings are my own and were done manually. Thank you!

[k1bac]

I must correct my previous findings, the error does occur with history enabled too. I seem to have failed to notice it and I do not have another workaround in mind.

[theotherp]

Please create and post debug infos.