That fixed it. Its working great in persistent mode on Red Hat 9 now. Thanks, Bill
-----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Pete McNeil Sent: Sunday, May 23, 2004 3:51 PM To: [EMAIL PROTECTED] Subject: RE: [sniffer] v2-3 persistent and Linux Importance: High At 01:16 AM 5/23/2004, you wrote: > > When you run it without & what happens? > >The same thing occurs...99% CPU utilization. I've figured it out. Attached are two updated source files that will bring you up to 2-3.0i1. The problem you were seeing is that clock_t values were overflowing when setting the timer for server persistence. This caused a race condition where the server instance would start, discover that it was out of time, exit, and repeat. This uncovered additional issues with the use of clock()... so an additional change was made to the CommandCheck() function to ensure that stop, reload, and rotate messages are processed in a timely manner. Copy sniffer.cpp and PeerServer.hpp into your source and recompile. That should solve your problem. I am going to expedite version 2-3.1 which will remove the use of clock(), unify all timing functions, and provide a "scoreboard" file in XML format that shows real-time performance data when the persistent server is running. This new version will eliminate all problems with clock() known or unknown. This appears to only effect some systems and conditions - so if anyone is not experiencing any problems then they should be fine with the current code until 2-3.1 is ready. Hope this helps, Thanks for your patience and support! _M This E-Mail came from the Message Sniffer mailing list. For information and (un)subscription instructions go to http://www.sortmonster.com/MessageSniffer/Help/Help.html
