Roll back fix for 1405440

[pecameron]

The fix to use a TCP connection check to check whether the HAProxy
process is alive or not doesn't work without a iptables rule for port
1936. The original test using HTTPGet works because HTTPGet supports
a Host field that can be set with "localhost" when host networking is
used. The TCPSocketAction does not support a Host field.

Rolling back the fix until a new fix is developed.

bug 1430729

Signed-off-by: Phil Cameron pcameron@redhat.com

[pecameron]

@knobunc @danwinship @dcbw Rolled back the fix for 1405440. PTAL

[danwinship]

standard git way to do this would be "git revert 5b708a5", which then makes it completely clear that this is just reverting an earlier commit

[pecameron]

@danwinship Did I get this right? PTAL

[danwinship]

yeah, looks right

[pecameron]

@danwinship @dcbw @eparis Is this ready to merge? If so will one of you merge it?

[louyihua]

Maybe the health check can be performed on port 80, which should be guarenteed to be permitted by iptables to support router's function. (#13345)
I'll also try to submit a PR on kubernetes so that the TCPSocketAction can have an option to use loopback address for probe. If this is implemented one day, we can switch to use the loopback TCPSocketAction on stats port again.
So @danwinship @dcbw @eparis what's your opinion?

[imcsk8]

It sounds like a good option

[pecameron]

@louyihua The HTTPGet handler also had problems (see comment pkg/cmd/admin/router/router.go line 423) so adding Host to the TCPSocketAction may be all that is needed.

[louyihua]

@pecameron
Yes, this is the best way to solve this problem, and I've prepared a preliminary PR submitted to kubernetes (kubernetes/kubernetes#42902) .

[eparis]

[merge] to fix broken, we can try again once everyone agrees on the next path.

[openshift-bot]

[Test]ing while waiting on the merge queue

[openshift-bot]

Evaluated for origin test up to 0d8009f

[openshift-bot]

continuous-integration/openshift-jenkins/test UNSTABLE (https://ci.openshift.redhat.com/jenkins/job/test_pull_request_origin/123/) (Base Commit: 93525f2)

[eparis]

[merge] but https://ci.openshift.redhat.com/jenkins/job/merge_pull_request_origin/45/testReport/junit/(root)/Extended/_Conformance__networking__router__openshift_routers_The_HAProxy_router_should_serve_the_correct_routes_when_scoped_to_a_single_namespace_and_label_set/ is a bit worrying that maybe it isn't a flake...

[eparis]

guess it was a flake [merge] because last time it was #11114

[openshift-bot]

Evaluated for origin merge up to 0d8009f

[openshift-bot]

continuous-integration/openshift-jenkins/merge SUCCESS (https://ci.openshift.redhat.com/jenkins/job/merge_pull_request_origin/54/) (Base Commit: 31e9bd0) (Image: devenv-rhel7_6067)