Loadbalancing issues with mod_jk 1.2.14
I tried to upgrade from 1.2.8 to 1.2.14 this week. I must say that the behavior of mod_jk has changed quite a bit. Especially troublesome is that the LB seems less equally distributing in 1.2.14 than it was in 1.2.8. Essentilly I experienced some Tomcats die under the load while others where bored. I know there are new settings in 1.2.14 for loadbalancing, namely lock and method. Can someone advise me which combination of flags would best recreate the loadbalancing as it was under 1.2.8? It is not that I want to stay on 1.2.8 but at this point I can't get 1.2.14 to run stable enough under high load. Thanks, Andreas - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Problem with mod_mpm / mod_jk / Tomcat Configuration
Under high load, the loadbalancing seems to forget session affinity and users have to log in again. Apache 2.0.54 with mod_mpm mod_jk 1.2.8 14 x Tomcat 4.1.31 instances The MaxClient setting seems to be causing most of these troubles. I don't exactly understand how the mpm and Tomcat settings are supposed to correlate. As soon as I increase MaxClients to 3200 the troubles start. mpm settings: IfModule worker.c StartServers 4 MaxClients2944 MinSpareThreads 25 MaxSpareThreads800 ThreadLimit 96 ThreadsPerChild 64 MaxRequestsPerChild 0 ListenBacklog 500 /IfModule Tomcat settings (same for the 14 instances): Connector className=org.apache.coyote.tomcat4.CoyoteConnector acceptCount=100 bufferSize=2048 compression=off connectionLinger=-1 connectionTimeout=6 connectionUploadTimeout=30 debug=0 disableUploadTimeout=false enableLookups=false maxKeepAliveRequests=500 maxProcessors=1000 minProcessors=50 address=%IP_ADDRESS% port=%CONNECTOR_PORT% protocolHandlerClassName=org.apache.jk.server.JkCoyoteHandler proxyPort=0 redirectPort=8443 scheme=http secure=false tcpNoDelay=false tomcatAuthentication=true useBodyEncodingForURI=true useURIValidationHack=false The max processors is set to 1000, I assumed that would result in 14000 processors for all instances and therefore the 3000 max clients of apache would be well within it. Is my assumption false adn I need to allocate 3000 thread per instance instead? Thanks, Andreas - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Problem with mpm/mod_jk/Tomcat Configuration
Under high load, the loadbalancing seems to forget session affinity and users have to log in again. Apache 2.0.54 with mod_mpm mod_jk 1.2.8 14 Tomcat 4.1.31 instances The MaxClient setting seems to be causing most of these troubles. I don't exactly understand how the mpm and Tomcat settings are supposed to correlate. As soon as I increase MaxClients to 3200 the troubles start. mpm settings: IfModule worker.c StartServers 4 MaxClients2944 MinSpareThreads 25 MaxSpareThreads800 ThreadLimit 96 ThreadsPerChild 64 MaxRequestsPerChild 0 ListenBacklog 500 /IfModule Tomcat settings (same for the 14 instances): Connector className=org.apache.coyote.tomcat4.CoyoteConnector acceptCount=100 bufferSize=2048 compression=off connectionLinger=-1 connectionTimeout=6 connectionUploadTimeout=30 debug=0 disableUploadTimeout=false enableLookups=false maxKeepAliveRequests=500 maxProcessors=1000 minProcessors=50 address=%IP_ADDRESS% port=%CONNECTOR_PORT% protocolHandlerClassName=org.apache.jk.server.JkCoyoteHandler proxyPort=0 redirectPort=8443 scheme=http secure=false tcpNoDelay=false tomcatAuthentication=true useBodyEncodingForURI=true useURIValidationHack=false The max processors is set to 1000, I assumed that would result in 14000 processors for all instances and therefore the 3000 max clients of apache would be well within it. Is my assumption false adn I need to allocate 3000 thread per instance instead? Thanks, Andreas - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
