Marcin Koziej wrote: > Hello! running resin 4.0.6. > > When running one web application everything works fine. But when I try to run > 50 separate instances of this app (this is a customer requirement), strange > things happen: > > Resin server stops responding, I get this kind of messages: > {http://*:80-5} TcpSocketLink[id=5,] failed keepalive (select) > and, after some time, messages like this: > slow alarm Alarm[alarm[ConnectionPool[jdbc/dataSource]]] 11087m > where jdbc/dataSource is a JNDI dataSource of course. > > I googled in vain. > How can I bite this problem?? > I'm working on the select manager at the moment.
As a workaround, you can disable the select manager with <select-manager-enable>false</select-manager-enable> in the cluster section. The slow alarm message should be unrelated, although it could be important. We added that log to check when alarms take too long to process. Normally, they should complete in milliseconds, so 11sec is slow. It could just mean a heavily loaded system or a long GC, but it could also be an indication of a threading problem, for example thread-max set too low with lots of connections, or an actual threading problem. At thread dump around that time might help see if there is an actual problem. -- Scott > Best regards, > > Marcin Koziej > > > > _______________________________________________ > resin-interest mailing list > resin-interest@caucho.com > http://maillist.caucho.com/mailman/listinfo/resin-interest > > _______________________________________________ resin-interest mailing list resin-interest@caucho.com http://maillist.caucho.com/mailman/listinfo/resin-interest