I’ve been trying to install rancher on an HA cluster scheme.
First i tried the HA air gap install
Second i setup a proxy on the servers and tried the normal HA install
Im installing behind a L7 LB (HA-proxy and KeepAlived) that terminates ssl for me, so i didn’t installed cert-manager and used --set tls=external on the installation according to documentation.
Everything works fine and all the pods are running. No events and the logs don’t show anything special to the exception of the rancher logs saying this lines:
2019/10/30 10:25:31 [ERROR] CatalogController library [catalog] failed with : Clone failed: Cloning into ‘management-state/catalog-cache/380859f1003fe7603cddc6c15b34b7263f1f0deaa92ddcde465811d032ee7078’…
fatal: unable to access ‘https://git.rancher.io/charts/’: Failed to connect to git.rancher.io port 443: Connection timed out
2019/10/30 10:24:38 [ERROR] ClusterController local [cluster-deploy] failed with : waiting for server-url setting to be set
Those errors seem normal cause the git project doesnt exist and i still havent logged into the website to choose the site url.
When i try to go directly to the rancher ui using the node i get this:
curl http://[node].[domain] -> default backend - 404
but the healthz gives me and http 200
curl http://[node].[domain]/healthz -v -> … < HTTP/1.1 200 OK …
That is what happens when i connect directly to the ingress.
If i used the DNS of the external load balancer:
https://[FQDN] -> infinite redirects to the root (/) path
https://[FQDN]/healthz -> error 400
I check the ingress logs and my request for /healthz goes there but doesnt seem to be forwarded to the rancher service
I’vent yet been able to see the rancher ui in any of my installations.
rancher version: 2.3.2
docker version: 18.09.9
kub version server v1.15.5: client: v1.12.9
helm version: 2.15.1
os: Redhat 7.7