2011/3/18 Mark Thomas <ma...@apache.org>: > On 18/03/2011 13:38, Francis GALIEGUE wrote: >> On Fri, Mar 18, 2011 at 11:46, Mark Thomas <ma...@apache.org> wrote: >>> My guess is that something is hanging somewhere, probably on some sort >>> of network activity - although I can;t think what it might be. When it >>> seems to be slow, take 2 or 3 thread dumps a couple of seconds apart. >>> That should point you towards the but of code causing the problem. >>> >> >> http://paste.pocoo.org/show/355534/ >> >> On a 30 seconds start sequence, I collected 3 traces and all look like >> this one. I don't see anything related to the manager webapp though :/ > > I do. The session ID generator is waiting for sufficient entropy to seed > the PRNG. The session ID generation in Tomcat 7 has been refactored for > speed (once it starts ;) ) and security. >
Just a note: Maybe it is worth to measure the time that the sessionId generation takes (looking at the stack trace in [1]) and print some warning if org.apache.catalina.util.SessionIdGenerator.getRandomBytes( ) or org.apache.catalina.util.SessionIdGenerator.generateSessionId( ) takes too long. I would say that more than 2 seconds would be too long. Just saying. [1] http://paste.pocoo.org/show/355534/ I see that ManagerBase.startInternal() already has debug logging around that call, but the OP does not have debug logging enabled. > You have a couple of options: > 1. Wait (if this is production I'd wait) > 2. Use a different secureRandomAlgorithm (see [1]) > 3. Use a different random source with: > JAVA_OPTS="-Djava.security.egd=file:/dev/./urandom" > in setenv.sh > > 3 will reduce the security of your session IDs. 2 will probably reduce > the security of your session IDs. > > Mark > > [1] http://tomcat.apache.org/tomcat-7.0-doc/config/manager.html > Best regards, Konstantin Kolinko --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org