Default UDP reporter protocol limits Jaeger production trace delivery

Jaeger agents default to reporting spans to collectors over UDP on port 6832. While this default is convenient for local development and flat networks, production environments with firewall rules, TLS proxies, or container network policies may block or drop UDP traffic. When that happens, traces can disappear silently without triggering application errors, leaving operators unable to tell whether the loss stems from sampling, network segmentation, or reporter configuration.

The unresolved decision is whether production deployments should treat the default UDP reporter as an unsupported assumption or whether the agent provides a supported path to detect and adapt to network constraints. Without a clear boundary, teams may discover trace gaps only after incidents.

Does the default UDP reporter protocol have documented production limitations in segmented or zero-trust networks? Should the agent reporter protocol be explicitly overridden when network policies block UDP, and if so, what configuration boundary applies? How can operators confirm that the agent's reporter protocol aligns with the collector's expectations without relying solely on observed trace volume?