[ https://issues.jboss.org/browse/MODCLUSTER-359?page=com.atlassian.jira.pl... ] Marco Danti updated MODCLUSTER-359: ----------------------------------- Fix Version/s: (was: 1.3.0.Alpha1) (was: 1.2.6.Final) Description: Setup: HTTPD configured as reverse proxy / load balancer for two instances of JBOSS-AS-7.2.0.Final running in HA configuration on two separate nodes. In the following configuration (defined inside my virtual hosts), the stickysession directive worked as expected using httpd-2.2.24/mod_cluster-1.2.0.Final. <Proxy balancer://mycluster> BalancerMember ajp://stw-1:7009/ route=stw-1 BalancerMember ajp://stw-2:7009/ route=stw-2 ProxySet stickysession=JSESSIONID|jsessionid </Proxy> After upgrading to httpd-2.4.6/mod_cluster-1.2.5.Final however, it does not work any longer and this is what happens: - if the 'BalancerMember' directives are omitted, then 'stickysession' is ignored and any client request gets dispatched alternatively to each one of the two JBoss servers (exactly as if lbmethod=byrequest were defined) - if the 'BalancerMember' directives are left in place, then the httpd server generates a segment fault in mod_proxy_cluster.c, in function "internal_find_best_byrequests" when attempting to access the variable 'helper' (defined as 'proxy_cluster_helper *helper') because the line helper = (proxy_cluster_helper *) worker->context; results in helper == NULL. As a final consideration, I read in the docs that stickysession should be enabled by default in mod_cluster anyway, but that is not the case for me. was: Setup: HTTPD configured as reverse proxy / load balancer for two instances of JBOSS-AS-7.2.0.Final on separate nodes. In the following configuration (defined inside my virtual hosts), the stickysession directive worked as expected using httpd-2.2.24/mod_cluster-1.2.0.Final. <Proxy balancer://mycluster> BalancerMember ajp://stw-1:7009/ route=stw-1 BalancerMember ajp://stw-2:7009/ route=stw-2 ProxySet stickysession=JSESSIONID|jsessionid </Proxy> After upgrading to httpd-2.4.6/mod_cluster-1.2.5.Final however, it does not work any longer: 'stickysession' is ignored, and any request gets dispatched alternatively to each one of my two JBOSS-AS-7.2.0 instances running in HA mode on two different nodes (exactly as if lbmethod=byrequest were defined) I read in the docs that stickysession should be enabled by default in mod_cluster anyway, but that is not the case for me. Please note that I get the same problem either with or without the two BalancerMember directives. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira