Add support for Ingress VIP #245
Conversation
|
Need to test the forwarders trick still |
|
Build FAILURE, see build http://10.8.144.11:8080/job/dev-tools/446/ |
|
It worked on my env. Should we keep the Ingress controller test? |
|
Ingress worked for me locally with this change. I like the test. It's kind of like the ping test from tripleo as a basic smoke test of the cluster. The only concern I have about keeping it is that I did have a spurious failure while running it. I haven't run this patch enough to know whether that will be a common problem though. I suppose we could merge it as-is and back out the test if it ends up being a problem (although if it is a problem we should probably address whatever is causing the underlying issue, so maybe that's an argument for keeping it too). |
|
+1 worked for me as well on my environment: [centos@provisionhost-0 dev-scripts]$ oc --config ocp/auth/kubeconfig get routes
NAME HOST/PORT PATH SERVICES PORT TERMINATION WILDCARD
tst-svc www.example.com tst-svc 8080 None
[centos@provisionhost-0 dev-scripts]$ oc --config ocp/auth/kubeconfig get svc
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
kubernetes ClusterIP 172.30.0.1 <none> 443/TCP 24m
openshift ExternalName <none> kubernetes.default.svc.cluster.local <none> 21m
tst-svc ClusterIP 172.30.81.253 <none> 1234/TCP 10m
[centos@provisionhost-0 dev-scripts]$ curl --header 'Host: www.example.com' whatever.apps.ostest.test.metalkube.org
tst-svc-1-h5ngb: HELLO! I AM ALIVE!!! |
There was a problem hiding this comment.
Overall code LGTM,
I run 6 deployments with the suggested changes in Libvirt environment.
Regarding the E2E routing test,
It passed 5/6, sometimes (2/5) it took ~600 seconds till the router/service respond (we should investigate that).
Test failed (1/6) with [1] error, I don't think it relates to routing functionality
[1]: + oc --config ocp/auth/kubeconfig run --image=kuryr/demo tst-svc
Error from server: rpc error: code = Unavailable desc = transport is closing
make: *** [ocp_run] Error 1
07_deploy_masters.sh
Outdated
| echo "Verifying Ingress controller functionality, first we'll check that router pod responsive" | ||
| until curl -o /dev/null -kLs --fail tst.apps.${CLUSTER_DOMAIN}:1936/healthz; do sleep 20 && echo "Router pod not responding"; done | ||
|
|
||
| echo "Create service named tst-svc" |
There was a problem hiding this comment.
There is already a lot of valid apps deployed in the cluster - like console - lets check that these are available?
if we want to keep the 'run arbitrary pod' scenario we'd have to take care of retries and error handling, which is terrible in bash
There was a problem hiding this comment.
@vrutkovs, done.
Changed the routing test to check connectivity to console service.
|
Build FAILURE, see build http://10.8.144.11:8080/job/dev-tools/456/ |
For the Openshift router support, the router pods will run on masters-nodes (host networking). This patch adds support for the Ingress VIP(192.168.111.4), one of the master nodes will hold this IP (VRRP), creates wildcard external DNS entry pointing to Ingress VIP and adds e2e verifictation test. Co-authored-by: Antoni Segura Puimedon <antoni@redhat.com>
|
Build SUCCESS, see build http://10.8.144.11:8080/job/dev-tools/462/ |
For the Openshift router support, the router pods will run on
masters-nodes (host networking).
This patch adds support for the Ingress VIP(192.168.111.4), one
of the master nodes will hold this IP (VRRP), creates wildcard
external DNS entry pointing to Ingress VIP and adds e2e verifictation test.
Co-authored-by: Antoni Segura Puimedon antoni@redhat.com