Issue Summary

When a customer's Jira DC instance returns a 504 Gateway Timeout on the journal events API (entity or permission), the Rovo connector falls into a retry storm — firing ~500+ HTTP requests/hour at the DC instance, well above the intended ~8 calls/hour. This can overwhelm the instance and trigger cascading node failures, effectively a self-inflicted denial-of-service.

We've seen this play out in production: a customer's 4-node DC instance hit 2000% CPU on individual nodes and went fully unresponsive. The only way to stop the traffic was for them to delete the Rovo application link.

Steps to Reproduce

1. Connecting Jira DC nodes with cloud using Rovo DC connector

Expected Results

Jira DC nodes shouldn't experience high CPU spikes under normal operation.

Actual Results

Jira DC nodes experience high CPU spikes under normal operation.

Workaround

Currently there is no known workaround for this behavior. A workaround will be added here when available