role of the various threads?
I'm trying to do some performance analysis. In my current test scenario I have a dedicated thread tomcat that is synchronously performing some work in a hard coded loop. For the purposes of timing/testing I have ensured that tomcat is not serving any requests while this some work is going on. Despite this, a quick run with runhprof shows that my thread is onlygetting a little under 10% of the cycles, and that tomcat threads called RMI - TCP Accept-0 RMI - TCP Accept-8333 main TP-Processor4 http-10722-Processor23 http-10722-Processor24 http-10722-Processor25 are each getting (roughly) as many ticks allocated as my thread. Any information as to what is going on, wether it's normal, where I can look for information on the various threads and types of thread that tomcat uses would be gratefully accepted. BugBear - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Re: role of the various threads?
paul womack wrote: RMI - TCP Accept-0 RMI - TCP Accept-8333 main TP-Processor4 http-10722-Processor23 http-10722-Processor24 http-10722-Processor25 are each getting (roughly) as many ticks allocated as my thread. Further reading has shown me that I'm using Tomcat5.5, and that the theads (above) are all running accept(). I thought accept() was blocking? BugBear - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Re: role of the various threads?
Caldarale, Charles R wrote: From: paul womack [mailto:pwom...@papermule.co.uk] Subject: Re: role of the various threads? Further reading has shown me that I'm using Tomcat5.5, and that the theads (above) are all running accept(). I thought accept() was blocking? Depends on the particular JRE being used on the particular platform. I would expect them to be blocking in most instances, but some esoteric systems use polling. You're getting down to a level that requires examination of not just Tomcat source code, but also JRE source and the platform's TCP/IP stack source - a bit beyond the normal scope of a support forum. It appears that hprof is counting blocked threads (since that *is* what the program is doing for some meanings of doing). So the threads are doing accept() but not actively consuming cycles whilst running accept(). Sigh. Thanks, hprof. BugBear - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Re: Tuning garbage collection
laredotornado wrote: Hi, I'm using Tomcat 6.0.26, Java 1.6 and wondering what tools/strategies you use to tune your garbage collection parameters? My main strategy is to see if I have any cripping GC problems. If not, I leave the GC to its own (or Sun's) devices. GC tuning is likely to be outdated/obseleted almost as soon as you've done it. BugBear - To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
Re: User-JDBC-Realm: User case-sensitiv?
Gregor Schneider wrote: Hi guys, I'm just a bit puzzled, maybe one of you can shed some light: We're running Tomcat 5.5 here having created a JDBC-realm holding our users their credentials: The DDL of the MySQL-Tables shows like However, we've just discovered that Tomcat doesn't care at all abot case-sensivity of the user-names being entered. I ran through the Tomcat-docs but couldn't find any hint on that issue. Needless to say that personally I find this behaviour quite annoying and would die for a hint how to get rid of that behaviour. I'm *guessing* that this could be a factor: http://dev.mysql.com/doc/refman/5.0/en/case-sensitivity.html BugBear - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Tomcat - JMX to get connections per minutes?
Caldarale, Charles R wrote: From: paul womack [mailto:[EMAIL PROTECTED] Subject: Tomcat - JMX to get connections per minutes? How do I get figures on (what I consider) simple stuff like connections per minute/hour/whatever, Kilobytes per second, etc? Probably the easiest thing to do is look at code that already does it. Lambda Probe (http://lambdaprobe.org) is a good open-source utility for that, as is MoSKito (http://moskito.anotheria.net/). Tomcat's own manager app also displays some statistics, and the source is in the src download. I was hoping that there already existed some (web?) or Swing tool that would graph/chart obvious performance/load parameters, analagous to sun PerfMeter, or KDE System Guard. I certainly wasn't intending to write a java app myself (unless forced) BugBear - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Tomcat - JMX to get connections per minutes?
Colour me stupid; I've installed Tomcat 5.5, and have read various glowing reports about monitoring and JMX and how wonderful it all is. I've just googled, and read stuff, for around the last hour. And come up empty. How do I get figures on (what I consider) simple stuff like connections per minute/hour/whatever, Kilobytes per second, etc? BugBear - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
undesirable session timeout behaviour
For reasons all my own, I want the timeout period to be quite short - essentially if a (human) user sits around doing nothing for more than 5 minutes, I want to timeout the session. I have configured tomcat to do this, and it works fine. Except. If the user initiates an activity (ok - I'll admit - it's a download) which takes longer than 5 minutes, the session gets timed out every time, since tomcat doesn't monitor activity in general for timeouts; tomcat just monitors user requests. I thought of setting the timeout to some large value at the start of a download, but that would require estimating bandwidth ahead of time (to estimate a good value for timeout). It would be more appealing to have download traffic reset the timeout clock, but I'm open to other insights. Has any body else had this problem, and (even better) created a solution? BugBear - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
role of threads - profiling?
I'm just trying to do some simple profiling on an app running under tomcat 4. I've enabled profiling using: JAVA_OPTS=-Xrunhprof:cpu=samples,depth=40,thread=y; export JAVA_OPTS and am successfully getting a java.hprof.txt file when tomcat is stopped. (after a 70 second run, with a perl script hammering away via http) On analysing the file with jperfanal http://jperfanal.sourceforge.net/JPerfAnal_manual.html I was intially completely confused. Then I realised than tomcat is (to say the least) multi-threaded, and went to use the tool's thread analyses. In terms of recorded ticks, the most important thread is main, with 9666 ticks, closely followed by: TP-processor4 (9470 ticks) After this I get a list of http10722-processorN threads, with ticks running down from 4580 - 487 all other threads have very few ticks. On examining function profiles for these threads, it is clear that the http threads are actually my application doing useful (IMHO) work. Can anyone outline to me what the main() and TP-processor threads are doing? I'd be happy to be pointed at relevant doc, but I've searched and read stuff for the last hour without really moving forward in my understanding. BugBear - To start a new topic, e-mail: users@tomcat.apache.org To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]