JBoss 7.2 was the basis for Redhat's EAP 6.1. Their supported stack is Apache 2.2.22 with mod_cluster 1.2.4. You will most likely get better results with those versions as it is a newer and tested configuration.
As for the error, if all the workers are in an error state it means apache thinks all of your JBoss instances are down and it can't proxy the request to JBoss. Check the mod_cluster-manager web page and make sure both of your JBoss instances along with the deployed webapps are registering with mod_cluster correctly and they are using the same balancer. You can enable debugging in apache to see the JBoss nodes registering with mod_cluster in Apache.