elektra
elektra copied to clipboard
[Elektra] Add alert for when nginx or oauth proxy returns 5xx error for longer than 10 minutes.
Alert for nginx ingress might be derived from the api availability metrics: https://plutono.global.cloud.sap/d/api-availability-overview/api-availability-overview?orgId=1&refresh=5m&var-period=5m&var-api=elektra
https://github.com/sapcc/helm-charts/blob/ff066bdff8c12dccf5ec6a8fe53ec09d97809f14/openstack/sre/templates/prometheus-aggregations.yaml#L33C1-L40
Oauth Proxy seems to have a metrics endpoint built in that is deactivated by default.