On Fri, Nov 28, 2003 at 12:23:53AM -0800, Stas Bekman wrote: > Jay R. Ashworth wrote: > > On Thu, Nov 27, 2003 at 05:12:10PM -0800, Stas Bekman wrote: > > > >>httpd -DONE_PROCESS -DNO_DETACH > > > > > > Oh, *ick*; it's a Heisenbug. > > > > This time it didn't dump core -- it did what it was doing originally: > > In which case you can run the server normally and still get the core > file, but in addition to the ulimit command you need to make the > directory where httpd resides writable by the user the server is > running under. The whole path to that directory needs to +x to that > user.
Apoligies; I wasn't clear: it didn't *segfault* this time; it went back to giving the mod_perl error (well, I think that's what it is). It segfaults if it's daemonized; it doesn't if it's standalone. > The easiest way for you to test whether you can get the core dumped is this. > let's say your httpd resides in /appl/Apache/bin/httpd, so we do: > > % su > % cd /appl/Apache/bin/ > % perl -MPOSIX -e 'my($uid, $gid) = (getpwnam(shift))[2..3]; \ > POSIX::setuid($uid); POSIX::setgid($gid); \ > print -r q{$dir} && -w _ && -x _ ? q{OK} : q{NOK};' \ > nobody > > If you get NOK that means that the core file won't be dumped. If OK, then > it'll work. In fact you can just try to dump the core as that user in a > similar way: > > % su > % cd /appl/Apache/bin/ > % perl -MPOSIX -e 'my($uid, $gid) = (getpwnam(shift))[2..3]; \ > POSIX::setuid($uid); POSIX::setgid($gid); \ > CORE::dump;' \ > nobody > > If you get: > > Abort > > That means that the core wasn't dumped. If you get: > > Abort (core dumped) > > then you will find a "nice" core file in /appl/Apache/bin/ I'll give this a try; thanks. -- j -- Jay R. Ashworth [EMAIL PROTECTED] Member of the Technical Staff Baylink RFC 2100 The Suncoast Freenet The Things I Think Tampa Bay, Florida http://baylink.pitas.com +1 727 647 1274 Come see Linux Gazette in our new home: www.linuxgazette.net! -- Reporting bugs: http://perl.apache.org/bugs/ Mail list info: http://perl.apache.org/maillist/modperl.html