I understand you're experiencing intermittent issues with your PDP where permission checks fail after running for about a week. The errors you're seeing indicate that the PDP server is unable to communicate with the internal OPA (Open Policy Agent) component (1) . Based on the error messages showing failures to reach

<http://localhost:8181>

, this appears to be an issue with the OPA service that runs inside the PDP container . The PDP uses OPA internally to evaluate policies, and when OPA becomes unresponsive or crashes, permission checks will fail . Here are some configuration changes that might help stabilize your deployment: 1. ​*Adjust OPA timeout settings*​ (2) : •

PDP_OPA_CLIENT_QUERY_TIMEOUT

- Default is 1 second, you might want to increase this •

PDP_OPA_CLIENT_FAILURE_THRESHOLD

- Default is 0.1 (10% failure rate triggers health check failure) •

PDP_OPA_CLIENT_FAILURE_THRESHOLD_INTERVAL

- Default is 60 seconds 2. ​*Monitor OPA health more closely*​ : • Enable

PDP_OPA_DECISION_LOG_CONSOLE=True

to see more detailed OPA logs • Set

OPAL_INLINE_OPA_LOG_FORMAT

to

full

or

http

for better debugging 3. ​*Consider resource limits*​ - The PDP might be running out of memory over time, causing OPA to crash (3) . To help our engineers investigate this issue further, could you please provide: 1. Which Workspace / Project / Env you are using 2. What specific SDK calls are you performing (as SDK code or CURL examples) 3. The exact PDP version (you mentioned v0.9.3 - can you confirm with

docker images

or your deployment manifest?) 4. The relevant PDP logs from when the issue occurs, including: • Logs showing the OPA connection errors • Any memory/CPU usage metrics from your monitoring • Logs from PDP startup showing the configuration being applied The logs you mentioned having from last Friday would be particularly helpful for diagnosis.

To learn more, see the sources I considered: [[1] Permit's PDP (Policy Decision Point)](https://docs.permit.io/concepts/pdp/overview) [[2] Configurations](https://docs.permit.io/concepts/pdp/configuration) [[3] How to properly scale out OPAL servers for data updates directly from Kafka](https://github.com/permitio/opal/discussions/671)