ID: 20270 Updated by: [EMAIL PROTECTED] Reported By: [EMAIL PROTECTED] -Status: Open +Status: Feedback Bug Type: Java related Operating System: RHL7.3 PHP Version: 4.3.0-pre2 New Comment:
Do you have xslt support enabled, then sablotron < 0.97 and jdk >= 1.3 does not work together. Sablotron 0.97 is not out yet, but there is an RC1 in their CVS (didn't find a link to download it), which should solve the problem.. chregu Previous Comments: ------------------------------------------------------------------------ [2002-11-22 13:11:07] [EMAIL PROTECTED] I get the same result using: Linux 2.4.19 glibc 2.3.1 pthread 0.10 php 4.3.0RC1 apache 1.3.27 Sun's j2sdk1.4.0_03 ------------------------------------------------------------------------ [2002-11-14 08:57:14] [EMAIL PROTECTED] There seems to be a problem with the way threads are handled between php and java that keeps the apache child process from cleaning up after itself. If MaxRequestsPerChild is set to anything other than 1, a number of threads are started by a httpd child and are never reused. It also appears that the httpd child process that spawns these threads can not be reused by apache and the httpd controlling process never starts another child process to take the place of the disabled one. My testing setup is: Linux 2.4.18 glibc 2.2.5 apache 1.3.27 libpthread 0.9 php 4.2.3 Blackdown JDK 1.3.1, IBM JDK 1.3..1, Blackdown JDK 1.4.1b2 all produce the same results. It should be noted that apache is stable if MaxRequestsPerChild is set to 1, but there's quite a performance hit involved. ------------------------------------------------------------------------ [2002-11-05 18:20:09] [EMAIL PROTECTED] I have tried to use the Java extension with JDK1.4.1 and PHP 4.2.3 and 4.3.0-pre2. Each time a page with embedded Java runs, the Apache (1.3.27) spawns MaxSpareServers processes, which never seem to get cleaned up. The tests I conducted with 4.2.3 ended up crashing after a few runs, being unable to run the JVM. This version also had trouble reading php.ini class path settings. The 4.3.0-pre2 seems to stop spawning processes after reaching about 90 httpds (not sure what this number is related to). The execution time of the code is much faster after the spawning stops also - probably the JVMs are loaded for all httpds? Is there a way to control this process spawning and to get the whole thing more stable? ------------------------------------------------------------------------ -- Edit this bug report at http://bugs.php.net/?id=20270&edit=1
