On Oct 22, 2011, at 8:51 PM, John Clingan <john.clingan_at_oracle.com> wrote:
>
> On Oct 22, 2011, at 5:20 PM, Tom Mueller wrote:
>
>> When you created the nodes and instances, did you do this from the Linux host or via the DAS?
>
> I did this from the Linux host using "create-node-config", "create-local-instance" and, "start-local-instnace".
>
>>
>> Can the DAS resolve t he Linux host's hostname? It sounds as though the DAS is unable to connect to the instances running on the Linux node.
>>
>
> Yes. I can ping (and ssh) in both directions. What is odd (to me) is that GMS thinks everything looks OK. I see all the GMS events properly showing up in the DAS server.log, and get-health shows the node is started.
GMS is using multicast over UDP, a different protocol stack.
Did you pass a --nodehost option to create-node-config? This should be the hostname for the Linux host that the DAS is able to resolve.
Tom
>
>> Tom
>>
>>
>>
>> On Oct 22, 2011, at 6:58 PM, John Clingan <john.clingan_at_oracle.com> wrote:
>>
>>> I've set up a cluster, with a DAS running on my MAC OS X host (JDK 1.6_26) and a second running in Oracle Linux (JDK 1.7.0.1) in a virtualbox instance (bridged networking).
>>>
>>> I successfully create nodes and instances on the Oracle Linux host, and they register fine with the DAS. When I tail the DAS server.log while starting the instances, everything looks OK. Instances join the cluster. asadmin get-health shows two instances have started.
>>>
>>> However, when I run "asadmin list-instances", it shows the instances as "not running". I've tried this with GlassFish 3.1.1 and GlassFish 3.1 (just in case). Same problem.
>>>
>>> Any thoughts?
>