> >But with the httpd.conf MaxRequestsPerChild parameter available to force
> >any and all memory leaks back to ground zero periodically, how bad *is*
> >this memory leak, compared to the number of people who find they can
> >no longer rotate their production server logs at modssl 2.6.5?
>
> In this case, it would be a memory leak in the parent, not in the child, since the
> problem happens when the server is shutting down and restarting. So having the
>children
> die off doesn't do any good because the new children get spawned with an image of
>the
> parent, leaked memory and all. Restart the server enough times, and your parent
>process
> gets to be pretty big in memory (and therefore its children do, too).
Ok, but here's an Apache modssl 2.5.3 parent that was started
yesterday morning at 9:33:
USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
root 17174 0.0 1.2 5792 2848 ? S 09:33:24 0:00 /www/apache/bin/ht
And here's one that has been running for over three months, restarted
nightly for log rotation:
USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
root 9445 0.0 1.0 5792 2344 ? S Apr 19 0:04 /www/apache/bin/ht
These are both on Ultra 2300's running Solaris 2.6.
Ed
--------------------------
Ed Kubaitis ([EMAIL PROTECTED])
CCSO - University of Illinois at Urbana-Champaign
______________________________________________________________________
Apache Interface to OpenSSL (mod_ssl) www.modssl.org
User Support Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
