E2E install used to check root ingress end to end with curl through root-ingress-controller at Host: grafana.example.org. #4023 turns off monitoring instance in tenant-root, so grafana is not installed anymore and this check goes away with it. It was the only assertion in the run that root ingress routes request to backend, rest of install waits on HelmReleases and Deployments, and they are Ready even if controller serves nothing.
Replacement should use host that install brings up anyway. Root tenant has seaweedfs enabled, it publishes s3.<root-host> and filer.<root-host>, so Host: s3.example.org is obvious candidate.
Open question is what response to match. Anonymous request to S3 gateway should return S3 XML (error or bucket listing), which is different from ingress-nginx default backend 404 and from 502/503 when service has no endpoints. I didn't check what seaweedfs returns here, it needs live stand before going into hack/e2e-install-cozystack.bats.
E2E install used to check root ingress end to end with
curlthroughroot-ingress-controlleratHost: grafana.example.org. #4023 turns off monitoring instance in tenant-root, so grafana is not installed anymore and this check goes away with it. It was the only assertion in the run that root ingress routes request to backend, rest of install waits on HelmReleases and Deployments, and they are Ready even if controller serves nothing.Replacement should use host that install brings up anyway. Root tenant has seaweedfs enabled, it publishes
s3.<root-host>andfiler.<root-host>, soHost: s3.example.orgis obvious candidate.Open question is what response to match. Anonymous request to S3 gateway should return S3 XML (error or bucket listing), which is different from ingress-nginx default backend 404 and from 502/503 when service has no endpoints. I didn't check what seaweedfs returns here, it needs live stand before going into
hack/e2e-install-cozystack.bats.