Zurab Kiknavelidze
08/12/2026, 11:11 AMRocky
08/12/2026, 11:11 AMstarrocks_be_query_rpc_latency: General RPC latency.
* starrocks_be_scan_rows / starrocks_be_scan_bytes: To track data volume being read.
* starrocks_fe_query_total: To see the total volume of queries (Arrow Flight queries are processed by the FE similar to MySQL protocol queries).
2. Service Availability
You can monitor whether the Arrow Flight service is active by checking the configured ports:
* BE Port: The arrow_flight_port (default is -1, meaning disabled) in be.conf.
* FE Port: The arrow_flight_sql_port in fe.conf.
* You can verify the service is up by checking if the BE/FE is listening on these ports or by inspecting the logs for [ARROW] Arrow Flight SQL server started.
3. Versions and Support
* v3.4.0+: Introduced the Arrow Flight SQL server on BE nodes.
* v3.5.1+: Introduced the Arrow Flight proxy on the FE, allowing clients to connect to the FE which then routes data from the BEs (useful for environments where BEs aren't directly accessible).
Summary Table
Resource | Monitoring Method
Service Status | Check arrow_flight_port in BE/FE config and port health.
Data Throughput | Monitor standard BE scan and transmission metrics.
Query Performance | Use the StarRocks Query Profile to see time spent in the "Result Deliver" phase.
If you are looking for specific metrics like arrow_flight_active_connections, they may exist internally but are not yet formally documented as public monitoring items. You can check your node's /metrics endpoint (typically http://<BE_IP>:8040/metrics) to see if any new arrow_flight strings appear in later versions.
References
• administration/management/monitoring/metrics.md