suggest using default cluster port on internal server error

[juanvallejo]

Related Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1406066

This patch suggests using the default cluster port with the current host
used by a client, if the current port does not match the default port,
and the client receives an internal / unknown error from the server.

Example

$ oc login --server=https://10.13.137.149:8444 -u system:admin
The server uses a certificate signed by an unknown authority.
You can bypass the certificate check, but any data you send to the server could be intercepted by others.
Use insecure connections? (y/n): y

error: Internal error occurred: unexpected response: 503
You may want to try using the default cluster port: https://10.13.137.149:8443

cc @openshift/cli-review

[juanvallejo]

[test]

[smarterclayton]

When does this happen?

[smarterclayton]

it's unlikely 503 means "use another port" in most setups. It's more likely to mean "all servers are down".

I would replace both lines with a message like "error: The server was unable to respond - verify you have provided the correct host and port and that the server is currently running."

[juanvallejo]

@smarterclayton

When does this happen?

It happens when a client attempts to connect to a non-default (non-8443) open port and receives an error, as well as when a client attempts to connect to a non-default closed port. We were already suggesting that users try using the default cluster port if the client attempted to connect to a closed port.

I would replace both lines with a message like "error: The server was unable to respond - verify you have provided the correct host and port and that the server is currently running."

Thanks, I have updated the error message to this one whenever a client attempts to connect to any open port and receives an error. Should this be the case as well for closed ports?

[smarterclayton]

I don't think we should suggest the default cluster port. No OpenShift installation I'm aware of except oc cluster up would use that port.

…

On Tue, Jan 24, 2017 at 5:37 PM, OpenShift Bot ***@***.***> wrote: continuous-integration/openshift-jenkins/test Running ( https://ci.openshift.redhat.com/jenkins/job/test_pr_origin/13269/) (Base Commit: 2ea8719 <2ea8719> ) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#12654 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABG_p7Kr2oFiMKgmj6QbUH-oP7aR10Qhks5rVn1DgaJpZM4LsxRk> .

[juanvallejo]

@smarterclayton

I don't think we should suggest the default cluster port. No OpenShift
installation I'm aware of except oc cluster up would use that port.

Okay, updated the last commit to return error: The server was unable to respond - verify you have provided the correct host and port and that the server is currently running. both when a client tries to log in to a closed port, and when it tries to log in to an open port but receives an internal / unknown error.

[juanvallejo]

conformance flaked on #11775 re[test]

[openshift-bot]

Evaluated for origin test up to 91b914e

[openshift-bot]

continuous-integration/openshift-jenkins/test SUCCESS (https://ci.openshift.redhat.com/jenkins/job/test_pr_origin/13347/) (Base Commit: 3e38a69)

[juanvallejo]

@smarterclayton are there any further comments on this patch? Thanks!

[juanvallejo]

cc @fabianofranz

[fabianofranz]

[merge]

[openshift-bot]

Evaluated for origin merge up to 91b914e

[openshift-bot]

continuous-integration/openshift-jenkins/merge SUCCESS (https://ci.openshift.redhat.com/jenkins/job/merge_pull_request_origin/21/) (Base Commit: d3a3181) (Image: devenv-rhel7_6061)