envoyproxy/envoy

How to distinguish externally and locally originated errors for upstream clusters?

Open

#25,798 opened on Feb 27, 2023

View on GitHub
 (1 comment) (0 reactions) (0 assignees)C++ (5,373 forks)batch import
area/clusterarea/statshelp wantedquestion

Repository metrics

Stars
 (27,997 stars)
PR merge metrics
 (Avg merge 8d) (378 merged PRs in 30d)

Description

Title: How to distinguish externally and locally originated errors for upstream clusters? Description:

In order to setup proper alerting for Api Gateway based on Envoy I'm trying to find the opportunity to get metric that shows only the errors originated by Envoy. The idea was to calculate the difference between downstream_rq_xx and upstream_rq_<*xx> but as I understand the upstream_rq_<*xx> includes also, for example, timeouts originated by Envoy.

Is there any stats for upstream cluster's externally originated errors? If no, could you please suggest any workaround to get such statistics?

[optional Relevant Links:]

I have found out that externally and locally originated errors are calculated separately during outlier detection: https://www.envoyproxy.io/docs/envoy/latest/intro/arch_overview/upstream/outlier But I have found any related statistics for upstream errors.

Contributor guide