Description of problem: When ServerThreshold is exceeded, it doesn't ignore the rest of the servers, it ignores the rest of the records including backend, frontend and server records. Version-Release number of selected component (if applicable): OCP 3.11 Actual results: Currently, OCP 3.11 does not have the backport fix of https://github.com/openshift/router/pull/12 made in 4.2 Expected results: OCP 3.11 to have this fix as well to not ignore the non-server records if the serverThreshold is reached. Additional info: Upstream 4.2 PR = https://github.com/openshift/router/pull/12 OCP 3.11 code reference = https://github.com/openshift/origin/blob/release-3.11/pkg/router/metrics/haproxy/haproxy.go#L428-L435
This is a straightforward backport.
verified with v3.11.154 and issue has been fixed. 1. oc set env dc/router ROUTER_METRICS_HAPROXY_SERVER_THRESHOLD=10 -n default 2. ensure all backend can be showed in the metrics even exceed the threshold. # curl admin:pass.0.1:1936/metrics <---snip---> # HELP haproxy_exporter_server_threshold Number of servers tracked and the current threshold value. # TYPE haproxy_exporter_server_threshold gauge haproxy_exporter_server_threshold{type="current"} 18 haproxy_exporter_server_threshold{type="limit"} 10
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2019:3817