BE Metrics

This document mainly introduces the monitor metrics of BE.

View Metrics

BE metrics can be viewed by visiting:

http://be_host:be_webserver_port/metrics

The default format is of Prometheus.

You can get Json format by visiting:

http://be_host:be_webserver_port/metrics?type=json

Metrics List

doris_be_snmp{name="tcp_in_errs"}

Value of the Tcp: InErrs field in /proc/net/snmp. Represents the number of error TCP packets currently received.

The incidence rate can be calculated in combination with the sampling period.

Usually used to troubleshoot network problems.

doris_be_snmp{name="tcp_retrans_segs"}

Value of the Tcp: RetransSegs field in /proc/net/snmp. Represents the number of error TCP packets currently received.

The incidence rate can be calculated in combination with the sampling period.

Usually used to troubleshoot network problems.

doris_be_snmp{name="tcp_in_segs"}

Value of the Tcp: InSegs field in /proc/net/snmp. Represents the number of receivied TCP packets.

Use (NEW_tcp_in_errs - OLD_tcp_in_errs) / (NEW_tcp_in_segs - OLD_tcp_in_segs) can calculate the error rate of received TCP packets.

Usually used to troubleshoot network problems.

doris_be_snmp{name="tcp_out_segs"}

Value of the Tcp: OutSegs field in /proc/net/snmp. Represents the number of send TCP packets with RST mark.

Use (NEW_tcp_retrans_segs - OLD_tcp_retrans_segs) / (NEW_tcp_out_segs - OLD_tcp_out_segs) can calculate the retrans rate of TCP packets.

Usually used to troubleshoot network problems.

doris_be_compaction_mem_current_consumption

The total MemPool consumption of all running Compaction threads. Use this value, we can easily identify whether Compactions use too much memory, it may cause memory overhead or OOM.

Usually used to troubleshoot memory problems.