multiple Apache instances with mod_cluster using common binaries

[dpalom]

Apache 2.2.15

mod_cluster 1.0.4 GA

JBoss EAP 5.1

I have a moderately large environment with multiple JBoss clusters.  Our general practice has been to use a one to one relationship between Apache instances and JBoss clusters.  Our multiple Apache instances share a single set of Apache binaries. 

I have found when trying to set up more than one Apache/mod_cluster/JBoss stack, mod_cluster seems to get confused.  mod_cluster-manager on any of the instances seems to know about the JBoss nodes that should be connected and visible only to other mod_cluster instances. 

It looks like the root of this behavior are the sockets/files that are created under <apache_bin>/logs/

cgisock.25704                       manager.domain.domain.lock     manager.node.nodes.slotmem

manager.balancer.balancers.lock     manager.domain.domain.slotmem  manager.sessionid.sessionid.lock

manager.balancer.balancers.slotmem  manager.host.hosts.lock        manager.sessionid.sessionid.slotmem

manager.context.contexts.lock       manager.host.hosts.slotmem

manager.context.contexts.slotmem    manager.node.nodes.lock

Is there a way to configure mod_cluster to tell it to put these objects in a specific location for each Apache instance, similar to how I already tell each Apache instance to put its access/error logs in a location specific to each instance? 

Thanks!

Dave

[jfclere]

See http://docs.jboss.org/mod_cluster/1.1.0/html/native.config.html#d0e553 (MemManagerFile)

[dpalom]

Strange.  That directive sure looks like what I was looking for, however it seems like it is being ignored.  FWIW, I am using mod_cluster 1.0.4 GA but the docs show that directive is available for 1.0.X also.  Maybe something else is wrong with my configs... 

Apache server root is /env/atlas.bomba.dev/cust/apache/usr/. 

The MemManager files are still showing up in /env/atlas.bomba.dev/cust/apache/usr/logs/ and not at the paths I am defining in each instance config. 

configs for apacheA:

LoadModule proxy_module modules/mod_proxy.so

LoadModule proxy_ajp_module modules/mod_proxy_ajp.so

LoadModule slotmem_module modules/mod_slotmem.so

LoadModule manager_module modules/mod_manager.so

LoadModule proxy_cluster_module modules/mod_proxy_cluster.so

LoadModule advertise_module modules/mod_advertise.so

<VirtualHost 10.199.128.222:18008>  

   <Directory />    

   Order deny,allow    

   Deny from all    

   #Allow from 10.199.128.

   Allow from all

</Directory>

KeepAliveTimeout 60

MaxKeepAliveRequests 0 

ManagerBalancerName eap-atlas0-LB

MemManagerFile /env/atlas.bomba.dev/cust/apache/http-eap-atlas-master-1/var/run/

AdvertiseFrequency 5

AdvertiseGroup 231.80.44.20

<Location /mod_cluster-manager>

SetHandler mod_cluster-manager

Order deny,allow

Deny from all

#Allow from 127.0.0.1

Allow from all

</Location>

</VirtualHost>

configs for apacheB:

LoadModule proxy_module modules/mod_proxy.so

LoadModule proxy_ajp_module modules/mod_proxy_ajp.so

LoadModule slotmem_module modules/mod_slotmem.so

LoadModule manager_module modules/mod_manager.so

LoadModule proxy_cluster_module modules/mod_proxy_cluster.so

LoadModule advertise_module modules/mod_advertise.so

<VirtualHost 10.199.128.222:18010>  

<Directory />    

   Order deny,allow    

   Deny from all    

   #Allow from 10.199.128.

   Allow from all

</Directory>

KeepAliveTimeout 60

MaxKeepAliveRequests 0 

ManagerBalancerName eap-atlas1-LB

MemManagerFile /env/atlas.bomba.dev/cust/apache/http-eap-atlas1-1/var/run/

AdvertiseFrequency 5

AdvertiseGroup 231.80.44.25

<Location /mod_cluster-manager>

SetHandler mod_cluster-manager

Order deny,allow

Deny from all

#Allow from 127.0.0.1

Allow from all

</Location>

</VirtualHost>

I know my Allows are wide open, just trying to get it to work before I ratchet that down. 

[jfclere]

Don't put it in the VIrtualHost. I think that the problem.

[dpalom]

Moving the directive out of the VirtualHost block did stop the files from showing up under <server_root>/logs but the behavior I see in mod_cluster-manager is the same. 

I updated the mod_cluster version to 1.0.10 GA and that seems to be working as I expected it to, no strange behavior.