Troubleshooting
Common startup and request-time failures across timestamping, with their causes and fixes.
| Symptom | Cause | Fix |
|---|---|---|
Startup fails: No suitable network address found for instance ID | PLATFORM_INSTANCE_ID is unset and no non-loopback IP is available | Set PLATFORM_INSTANCE_ID to any value 0–65,535 |
Startup fails: IllegalArgumentException on PLATFORM_INSTANCE_ID | Value is not an integer, or is outside 0–65,535 | Correct the environment variable |
| TSP requests rejected with a time not available error | Clock moved backward by more than 100 ms | Re-synchronize the system clock via NTP; avoid large manual clock adjustments on a running instance |
| TSP requests rejected with a time not available error | System clock is set before 2026-02-01T00:00:00Z | Correct the system clock |
| TSP requests rejected with a time not available error | The Time Quality Monitor is not reporting results at all, or not reporting them frequently enough, so the stored result has gone stale | Check that the Time Quality Monitor is running and successfully reporting; check that the NTP check interval is shorter than the Accuracy value — recommended at roughly half of Accuracy. See Time Quality Configuration |
| Brief latency spikes under burst load | 256 serials consumed in one 10 ms tick; generator spin-waiting for the next tick | Expected at very high throughput; scale horizontally if sustained above ~25,600 tokens/sec per instance |
| Duplicate serial numbers in audit logs | Two replicas share the same instance ID | Assign a distinct PLATFORM_INSTANCE_ID to each replica and restart; audit the affected time range for tokens that may need to be reissued |
Related pages
- Serial number generator — instance ID configuration and serial number generation
- Time Quality Configuration — Accuracy and NTP check interval settings
- Time Quality Monitor — how NTP check results are produced and reported