Just to double-check, does the camel-spring or camel-cdi quickstart work alright for you?

Tried reproducing using the steps provided, works here.  Can you make sure you've cleaned your browser cache just to be on the safe side?  Also, while in the hawtio console you can clear your local storage, click on the 'User' menu at the top-right, select 'Preferences' and then select the 'Reset' tab on the preferences page, click the (I think) red button to clear your local storage too.

@Stan Lewis, I've tried with your suggestion, still got 503 error in OSE env, but on fodera instance does not have this issue.  Could you please try on OSE env ?

@Stan Lewis, tested by camel-spring, activemq, java-simple-mainclass.

@zhou ying, any docs you can link me to for setting that up?  Thanks!

Also, could it be this PR hasn't made it into the build you're testing -> https://github.com/openshift/origin/pull/2802 as that's the behavior seen before that fix was put in.

Commit pushed to master at https://github.com/openshift/origin

https://github.com/openshift/origin/commit/641a56b7250bc5350f74877fd5859d3699b09439
Updates to address part of bug 1230483

Update openshift-jvm to 1.0.20

Change icon/title attribute and move to the right place

Remove absolute positioning

Ah, the above should have been for 1231605, sorry

K, have a VM with RHEL 7 and the openshift RPMs you linked me to.  Installed the java-simple-mainclass quickstart, clicked the connect button and the console was able to successfully establish a connection to Jolokia.  So I don't seem to be able to reproduce it locally still.

Could send along a disk image of what you've got setup and I can try running it here so we're on the same page...

@Stan Lewis, I just send email to you about  today's OSE env , you can use it to reproduce this issue.

Response from proxy ->

Error: 'dial tcp 10.1.0.40:8778: i/o timeout'
Trying to reach: 'http://10.1.0.40:8778/jolokia/?maxDepth=7&maxCollectionSize=500&ignoreErrors=true&canonicalNaming=false'

From master.cluster.local directly trying to curl:

curl "http://10.1.0.40:8778/jolokia/?maxDepth=7&maxCollectionSize=500&ignoreErrors=true&canonicalNaming=false"
curl: (7) Failed to connect to 10.1.0.40 port 8778: Connection timed out

Investigation showed the environment didn't have the master on the same network (sdn) as the nodes.

The 503 error only occurred when the env was multi-nodes, the reason is that on node there was vxlan ,thus the pods on different node can connect with each others.
But on master machine, there was not vxlan exist, so on the master machine can not connect the pod deployed on node. For the purpose of access pod from master, we should use service and route. I've tried by the json:http://fpaste.org/233414/43460686/, to create pod,service,route. Then I use command curl on master can get the response, please see the details: http://fpaste.org/233415/07068143/

@Stan Lewis, When I add service and route for pod, on master I can use curl to get response, but on Webconsole, the 503 error still exist. So I think maybe the connect url need to update. 

Here is doc about route: https://github.com/openshift/origin/blob/master/docs/routing.md

This would take a bit more work to support this use-case, as I suspect we'd have to define a reasonable annotation on the service that the console would need to use for discovery.  Work would also require relating the pod to the service then as well.  That being said it also then requires exposing the jolokia port for each pod to the outside, which I suspect most users would not want to do really as a safety precaution, even with jolokia secured.

Xiaoli,

The way we're going to address this for now is to make the master also be a node however make it unschedulable so it doesn't get any pods. In the future we'll make the master join the SDN without the additional overhead of running the node components. So for testing purposes provision the master as if it were a node and then make it unschedulable. We're working to automate this via ansible.

# oadm manage-node ose3-master.example.com --schedulable=false

Created attachment 1041579 [details]
console log

Created attachment 1041580 [details]
console log

(In reply to chunchen from comment #19)

After make the master as a node, can get below messages:

[root@master ~]# ps -ef |grep node
root       6965      1  3 11:33 ?        00:02:14 /usr/bin/openshift start node --config=/etc/openshift/node/node-config.yaml --loglevel=4

1. For schedulable=false:
[root@master ~]# oc get node
NAME                    LABELS                                         STATUS
master.cluster.local    kubernetes.io/hostname=master.cluster.local    Ready,SchedulingDisabled
minion1.cluster.local   kubernetes.io/hostname=minion1.cluster.local   Ready
minion2.cluster.local   kubernetes.io/hostname=minion2.cluster.local   Ready

2. For schedulable=true:
[root@master ~]# oc get node
NAME                    LABELS                                         STATUS
master.cluster.local    kubernetes.io/hostname=master.cluster.local    Ready
minion1.cluster.local   kubernetes.io/hostname=minion1.cluster.local   Ready
minion2.cluster.local   kubernetes.io/hostname=minion2.cluster.local   Ready

Created attachment 1041610 [details]
OK for java console

Also confirmed on the env according to https://trello.com/c/KtD20Ei8/341-2-admin-can-install-openshift-using-a-python-wrapper-around-the-ansible-tooling-with-no-additional-dependencies-beyond-base-rhel.

[root@master ~]# oc get nodes
NAME                   LABELS                                        STATUS
master.cluster.local   kubernetes.io/hostname=master.cluster.local   Ready,SchedulingDisabled
minion.cluster.local   kubernetes.io/hostname=minion.cluster.local   Ready

openshift version
openshift v3.0.0.0-32-g3ae1d27
kubernetes v0.17.1-804-g496be63

*** Bug 1243317 has been marked as a duplicate of this bug. ***