Greetings all. I would appreciate some advise on an issue I am experiencing. The problem is as follows: Firstly, lets set the stage: Server: Windows 2000 Server 1GB RAM Tomcat 5.0.19 Apache 2.0.46 mod_jk2 My configurations are: 40 VirtualHosts in Apache with the matching 40 Tomcat Hosts with 1 Context per Tomcat host. Each Apache VirtualHost uses the JkUriSet to map JSP and relevant servlets through to an AJP1.3 worker in Tomcat. Now, the problem is that periodically the sites stop responding due to the AJP Threadpool reaching its maximum limit. Now I could put the maxThread attribute up and still use one AJP connector accross all sites - but as additional sites (Hosts) get added to the server I would need to increase this (periodically). Now - the other option I have is to add a new AJP connector/worker for EACH site (host). This however could cause too many concurrent connections to be "allowed" and thus increase the memory requirement for the server (which is not a major problem). I am unsure (hence this post). Essentially I am trying to figure out a good balance for this problem. Can anyone tell me how JK2 works from a connection level. I know that it uses a pool - but how does this actually function. If I have 40 virtual hosts that use the same worker/connector when are the connections established, and how long do they stay open for? I did a netstat on my server and found over 167 established connections to port 8009 (the AJP Connector port). Do they close? Well, this does not seem to happening as they server does "fall over" when the maximum connection limit is reached. My AJP/1.3 connector looks as follows in my server.xml: <Connector port="8009" protocol="AJP/1.3" maxPostSize="0" maxThreads="500" minSpareThreads="20" acceptCount="30" /> Do the maxThreads and minSpareThreads and acceptCount work for the AJP/1.3 Connector? In the TC docs they are not included there? Maybe I am just confused. Any assistance here would be greatly appreciated! Best regards, Carl
