On 08/04/2011 09:58, Tim Whittington wrote: > The Http11Protocol and Http11NioProtocol connectors set processorCache > to 200 by default, which matches the docs ([2]). > The Http11AprProtocol sets it to -1 (unlimited) - is this > intentional/desired or accidental? > > This appears to have been introduced in [1] during some refactoring by Mladen.
processorCache was -1 for HTTP APR prior to this. Digging through svn shows that it has always had an unlimited cache. AJP (both BIO and APR) uses an unlimited cache (although the docs say 200). With the polling nature of APR and NIO and also with async connections it is certainly possible that more processors will be required than threads. More so with NIO that doesn't simulate blocking while the request line is read. However, that doesn't necessarily mean that there will be more than 200 idle processors. 200 seems like a reasonable default at the moment. Since the documentation everywhere says the default is 200, I will make that change shortly along with a better description of what an appropriate setting is. Mark > [1] http://svn.apache.org/viewvc?view=revision&revision=991359 > [2] http://tomcat.apache.org/tomcat-7.0-doc/config/http.html > > cheers > tim > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org > For additional commands, e-mail: dev-h...@tomcat.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
