First, many thanks to the only respondent, Casper Dik. As soon as I did what he suggested, I was able to determine the exact cause of the problem. Second, a brief description of the problem: When attempting to start Apache from the /etc/init.d/apache startup script, the apache startup script would hang at the following line: status=`${APACHE_HOME}/bin/apachectl $1 2>&1` Using 'truss' I had determined that the script was attempting a read on what should have been a closed pipe that was opened when the apache script forked for the `apachectl` command. Since the process that was forked for the `apachectl` command was shown as 'defunct', I couldn't figure out why the read on what was supposed to be a closed pipe did not return EOF. Thirdly, the solution: Per Casper's recommendation, I ran the following: pfiles apache_PID which yielded the following: [ Unimportant stuff cut out ] 3: S_IFIFO mode:0000 dev:228,0 ino:144 uid:0 gid:1 size:0 O_RDWR indicating that someone somewhere still had the file descriptor open. Executing, again per Casper's recommendation, the following: find /proc -type p -inum 144 yielded: /proc/487/fd/1 /proc/487/fd/2 /proc/482/fd/3 which shows that pid 487 still has the pipe open, which is why the read by pid 482 didn't return EOF. Further investigation showed that the offending pid was a log rotate script added to the Apache startup process by our development team in the last build, which explains why this problem just started to appear. Finally, many thanks to Casper for his help and providing a little education. I learned several valuable lessons as a result of this. Todd Urie _______________________________________________ sunmanagers mailing list sunmanagers@sunmanagers.org http://www.sunmanagers.org/mailman/listinfo/sunmanagersReceived on Tue Sep 17 08:33:28 2002
This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:42:55 EST