Latency or errors increase

Triage

1. Establish the start time and affected operations.
2. Compare traffic through Nexo with a safe baseline.
3. Check runtime health, restarts, resource pressure, and connection counts.
4. Review recent software and pipeline changes.
5. Check network latency and packet loss on both connection legs.
6. Inspect MongoDB health and operation latency.
7. Disable or roll back the most recent pipeline revision in a controlled environment.

Correlate client, Nexo, network, and MongoDB timestamps. A proxy-visible delay does not by itself identify the source. Scale or tune only after measurement identifies the constrained resource.