Modcluster finding other Nodes

[isa154]

Hi All

I am using Wildfly-8.0.0-Final with modcluster. I had this setup working fine in my INT  environment. Today I started moving to QA and when I started modcluster, it found my int environments contexts which I can confirm by looking at mod_cluster_manager pager. Also my INT environment picked up my QA contexts.  How do I stop this from happening ?

Here is my Modcluster  (1.1.1-2.el6) settings:

#

# This is mod_cluster module configuration file. Please refer to README

# for more information on how to enable mod_cluster.

LoadModule proxy_module         modules/mod_proxy.so

LoadModule slotmem_module       modules/mod_slotmem.so

LoadModule proxy_ajp_module     modules/mod_proxy_ajp.so

LoadModule manager_module       modules/mod_manager.so

LoadModule proxy_cluster_module modules/mod_proxy_cluster.so

LoadModule advertise_module     modules/mod_advertise.so

CreateBalancers 1

<IfModule manager_module>

   ListenHOSTNAME:9000

  <VirtualHost HOSTNAME:9000>

    #Services

    ProxyPass /AccessControl/ balancer://mycluster/AccessControl/ nofailover=On

    ProxyPassReverse /AccessControl/ balancer://mycluster/AccessControl/

    ProxyPreserveHost On

    <Location /mod_cluster_manager>

       SetHandler mod_cluster-manager

    </Location>

    ManagerBalancerName mycluster

        KeepAliveTimeout 300

        MaxKeepAliveRequests 0

        AdvertiseFrequency 5

        ServerAdvertise On

        #EnableMCPMReceive

    #Log

        ErrorLog "logs/mod_cluster_error_log"

        CustomLog "logs/mod_cluster_access_log" common

        LogLevel info

</VirtualHost>

</IfModule>

Here is my standalone.xml:

<mod-cluster-config advertise-socket="modcluster" proxy-list="HOSTNAME:9000" proxy-url="/" advertise="true" excluded-contexts="ROOT,invoker,jbossws,juddi,console" auto-enable-contexts="true" stop-context-timeout="10" socket-timeout="20" sticky-session="false" sticky-session-remove="false" sticky-session-force="false" worker-timeout="-1" max-attempts="1" flush-packets="false" flush-wait="-1" ping="10" smax="-1" ttl="-1" node-timeout="-1" connector="ajp"/>

[rhusar]

So the problem ist that the QE deployemnts registered with the production environment? You need to separate them cleanly.

There are lot of ways to do this:

• use security advertize key
• use different  multicast group (AdvertiseGroup)
• use allow from/deny from to filter for correct IP address ranges
• disable server advertise (ServerAdvertise off) and configure proxy-list (in WildFly configuration)

Honestly, the first 3 should have been already done as a security precaution.

[isa154]

No the Problem was between QA and INT  not Production. The next step would be production

what worked for me is

disable server advertise (ServerAdvertise off) and configure proxy-list (in WildFly configuration)

I will have a look at the others.

use different  multicast group (AdvertiseGroup)

Is this the multicast address under socket binding?

<socket-binding name="modcluster" port="0" multicast-address="224.0.1.105" multicast-port="23364"/>

[rhusar]

Is this the multicast address under socket binding?

<socket-binding name="modcluster" port="0" multicast-address="224.0.1.105" multicast-port="23364"/>

Yes, but note that both parts needs to use the same multicast group and address. See the docs mod_cluster Documentation

So to efficiently split them, both test AS and test Apache couple need the same address and then prod AS and prod Apache needs to use same but different from the first couple.

[isa154]

both test AS and test Apache couple need the same address

Last question, so how do I add this this address to apache?

Thanks for the help so far, you always very helpful

[isa154]

Hi

use security advertize key

I have implemented the above but it does not seem to work. I have set a key in WF and a different key in httpd but the nodes still register on mod_cluster.  The reason I am trying this is because

disable server advertise (ServerAdvertise off) and configure proxy-list (in WildFly configuration)

actually did not work for me

[rhusar]

Hm, strage it doesn't work, there is also test in the testsuite that validates this logic. What versions are you using?

If you dont configure one in the wildfly subsystem, it will not be validated (this has changed in 1.3.x mocluster and wildfly 8 iirc).

Take a look at the above comment (Re: Modcluster finding other Nodes) and implement a different way.