[jboss-as7-dev] TCPPING in AS7

Wed Mar 14 20:29:52 EDT 2012

There is no multicast - it's not supported in EC2. It's all point to 
point TCP. Thus TCPPING. The was working in AS7.0.x. Maybe there's an 
iptables issue that's preventing the 2 from seeing each other. Looking 
into other causes.

Thanks -Bill

On 3/14/12 5:25 PM, Richard Achmatowicz wrote:
> Are you sure your multicast messages are getting through? Bela has a 
> utility called McastSendTest and McastReeceiverTest in the JGroups 
> distro which allows checking if the network is set up OK for multicast.
>
> On 03/14/2012 08:21 PM, William DeCoste wrote:
>> OK, that's what I thought - the 2 nodes aren't seeing each other. 
>> Thanks for confirming the config is ok. I never see any logging with 
>> a membership of more than one. And I never saw any change if I killed 
>> a node.
>>
>> On 3/14/12 5:18 PM, Richard Achmatowicz wrote:
>>> On 03/14/2012 08:15 PM, William DeCoste wrote:
>>>> Hi Richard,
>>>>
>>>> I'm also deploying a clustered web app which causes JGroups and the 
>>>> rest of clustering to load.
>>>>
>>>> Maybe this is just a change in logging. I am seeing the following 
>>>> with AS7.1.0:
>>>> 2012/03/14 19:53:21,710 INFO  
>>>> [org.infinispan.remoting.transport.jgroups.JGroupsTransport] 
>>>> (pool-17-thread-1) ISPN000094: Received new cluster view: 
>>>> [ip-10-190-239-128/ejb|0] [ip-10-190-239-128/ejb]
>>>>
>>>> Does "[ip-10-190-239-128/ejb|0] [ip-10-190-239-128/ejb]" indicate 2 
>>>> nodes? This is on Express thus the strange node names.
>>> One node. IIRC, the first element [ip-10-190-239-128/ejb|0] 
>>> indicates the host which triggered the view change, and the second 
>>> element [ip-10-190-239-128/ejb] is the new group membership.
>>>>
>>>> I'm used to seeing jgroups debug logging of the pings that isn't 
>>>> happening. Was definietly seeing different logging in 7.0.x.
>>>>
>>>> Thanks -Bill
>>>>
>>>> On 3/14/12 4:55 PM, Richard Achmatowicz wrote:
>>>>> Hi Bill
>>>>>
>>>>> The configuration is valid. I tried it out on my local machine. I 
>>>>> set up two AS 7.1.1. instances, configured as you describe but 
>>>>> using a different set of IPv4 addresses (I used 192.168.0.103, 
>>>>> 192.168.0.104 and used "iptables -F" to let multicast messages 
>>>>> through).
>>>>>
>>>>> Because clustering services are installed on-demand, they need 
>>>>> some event to to start them, so I also deployed a small clustered 
>>>>> web application.
>>>>>
>>>>> When I start both instances, no channels are started, as expected:
>>>>> $ ./standalone.sh --server-config standalone-ha.xml 
>>>>> -Djboss.bind.address=192.168.0.103 
>>>>> -Djboss.bind.address.management=192.168.0.103 -Djboss.node.name=A
>>>>> $ ./standalone.sh --server-config standalone-ha.xml 
>>>>> -Djboss.bind.address=192.168.0.104 
>>>>> -Djboss.bind.address.management=192.168.0.104 -Djboss.node.name=B
>>>>>
>>>>> When I deployed the clustered web app on each application server 
>>>>> instance to trigger Infinispan web cache container and JGroups tcp 
>>>>> channel startup, I saw for example, on the first host A:
>>>>>
>>>>> 18:58:41,086 INFO  [org.jboss.as.clustering.infinispan] (MSC 
>>>>> service thread 1-8) JBAS010281: Started 
>>>>> //default-host//my-clustered-webapp cache from web container
>>>>> 18:58:41,096 INFO  
>>>>> [org.infinispan.configuration.cache.EvictionConfigurationBuilder] 
>>>>> (MSC service thread 1-8) ISPN000152: Passivation configured 
>>>>> without an eviction policy being selected. Only manually evicted 
>>>>> entities will be pasivated.
>>>>> 18:58:41,099 INFO  
>>>>> [org.infinispan.configuration.cache.EvictionConfigurationBuilder] 
>>>>> (MSC service thread 1-8) ISPN000152: Passivation configured 
>>>>> without an eviction policy being selected. Only manually evicted 
>>>>> entities will be pasivated.
>>>>> 18:58:41,220 INFO  [org.jboss.web] (MSC service thread 1-8) 
>>>>> JBAS018210: Registering web context: /my-clustered-webapp
>>>>> 18:58:41,230 INFO  [org.jboss.as.server] (Controller Boot Thread) 
>>>>> JBAS018559: Deployed "my-clustered-webapp.war"
>>>>> 18:58:41,333 INFO  [org.jboss.as] (Controller Boot Thread) 
>>>>> JBAS015951: Admin console listening on http://192.168.0.103:9990
>>>>> 18:58:41,334 INFO  [org.jboss.as] (Controller Boot Thread) 
>>>>> JBAS015874: JBoss AS 7.1.1.Final-SNAPSHOT "Thunder" started in 
>>>>> 6719ms - Started 175 of 305 services (129 services are passive or 
>>>>> on-demand)
>>>>> 18:58:50,673 INFO  
>>>>> [org.jboss.as.clustering.impl.CoreGroupCommunicationService.lifecycle.web] 
>>>>> (Incoming-1,null) JBAS010247: New cluster view for partition web 
>>>>> (id: 1, delta: 1, merge: false) : [A/web, B/web]
>>>>> 18:58:50,687 INFO  
>>>>> [org.infinispan.remoting.transport.jgroups.JGroupsTransport] 
>>>>> (Incoming-1,null) ISPN000094: Received new cluster view: [A/web|1] 
>>>>> [A/web, B/web]
>>>>>
>>>>> So, it seems as though things work as expected.
>>>>>
>>>>> I also tried to achieve the same effect by not deploying the web 
>>>>> app and setting the attribute start="EAGER" on a clustered 
>>>>> Infinispan cache (in this case, cache "repl" in the "web" cache 
>>>>> container) to force them to start as active services. When the 
>>>>> application servers were started this time, same result: 
>>>>> clustering between the two nodes was visible.
>>>>>
>>>>> If I instead set the start="EAGER" attribute on a cache container, 
>>>>> however, the channels do not get started automatically and I need 
>>>>> to deploy a web app as before to trigger starting of the transport.
>>>>>
>>>>> So, are you using start=EAGER to start the cache instances, or 
>>>>> deploying an app to trigger them? The organization of Infinispan 
>>>>> services has changed a lot since 7.1.0. In particular, a cache 
>>>>> container needs to specify the <transport/> element in order for a 
>>>>> transport to be used.
>>>>>
>>>>> Richard
>>>>>
>>>>> On 03/14/2012 05:47 PM, Brian Stansberry wrote:
>>>>>> Hi Paul, Richard:
>>>>>>
>>>>>> Do you guys see anything wrong here?
>>>>>>
>>>>>> - Brian
>>>>>>
>>>>>> On 3/14/12 4:44 PM, William DeCoste wrote:
>>>>>>> Hi Brian,
>>>>>>>
>>>>>>> Has anything changed in the config of AS7.1 from 7.0.x? I am not 
>>>>>>> seeing
>>>>>>> any jgroups traffic or discovery. This configuration worked fine in
>>>>>>> 7.0.x. When JGroups is loaded it just creates 2 1-node clusters. 
>>>>>>> They
>>>>>>> don't seem to see each other or even be trying.
>>>>>>>
>>>>>>> Thanks -Bill
>>>>>>>
>>>>>>> <subsystem xmlns="urn:jboss:domain:jgroups:1.1" 
>>>>>>> default-stack="tcp">
>>>>>>> <stack name="tcp">
>>>>>>> <transport type="TCP" socket-binding="jgroups-tcp"/>
>>>>>>> <protocol type="TCPPING">
>>>>>>> <property name="timeout">
>>>>>>> 3000
>>>>>>> </property>
>>>>>>> <property name="initial_hosts">
>>>>>>> 127.0.250.1[7600],127.0.251.1[7600]
>>>>>>> </property>
>>>>>>> <property name="port_range">
>>>>>>> 1
>>>>>>> </property>
>>>>>>> <property name="num_initial_members">
>>>>>>> 2
>>>>>>> </property>
>>>>>>> </protocol>
>>>>>>> <protocol type="MERGE2"/>
>>>>>>> <protocol type="FD_SOCK" socket-binding="jgroups-tcp-fd"/>
>>>>>>> <protocol type="FD"/>
>>>>>>> <protocol type="VERIFY_SUSPECT"/>
>>>>>>> <protocol type="BARRIER"/>
>>>>>>> <protocol type="pbcast.NAKACK"/>
>>>>>>> <protocol type="UNICAST2"/>
>>>>>>> <protocol type="pbcast.STABLE"/>
>>>>>>> <protocol type="pbcast.GMS"/>
>>>>>>> <protocol type="UFC"/>
>>>>>>> <protocol type="MFC"/>
>>>>>>> <protocol type="FRAG2"/>
>>>>>>> </stack>
>>>>>>> </subsystem>
>>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>

-- 
Bill DeCoste
Principal Software Engineer, Red Hat
978-204-0920
wdecoste at redhat.com