[imp] Apache Random Segmentation Faults with IMP

Mike Cochrane mike@graftonhall.co.nz
Thu, 15 Aug 2002 07:10:14 +1200


On redhat 7.3, my solution was to drop the mcrypt module and it was all fine
after that..

- Mike :-)

Quoting Jeremy Freeman <jfreeman@sporg.com>:

> Here's something I have discovered: Not sure if it will pertain to
> others however. If I compile PHP with --enable-debug AND _not_ using the
> included mysql libraries the problem seems manageable. By manageable I
> mean it STILL segfaults frequently, however the segfault does not cause
> the browser to display a "Page cannot be displayed" message. So the
> problem has become transparent to the users. I would venture to guess
> that the timing of the segfault has been adjusted slightly so it occurs
> after the page content is delievered to the browser.
> 
> --
> Jeremy
> 
> Craig A Lewis wrote:
> > Hello,
> >
> > I have the same problem. I too have reported it to the imp
> > group. Whats slightly interesting is that we all have
> > fairly different systems I think. Ours is solaris, and I
> > can't tell for sure about you guys, but we have ssl going
> > between browser and apache as well as IMAPS between apache/php
> > and the UW IMAP server daemon. What I can't figure out is why
> > I am not getting the Seg faults from apache. I have looked
> > carefully at my LogLevels, etc. but I just don't understand
> > why I am not getting the Segfault error, because even the
> > konqueror browser is reporting that the https daemon is
> > dying. I can excite the bug much more frequently. I can get
> > the compose window, send mail sequence to fail 50-80% of
> > the time. My latest theory was that there is an nfs lock
> > problem of some sort, but best I can tell, all 'scratch'
> > session files are in /tmp which is a local disk, and while
> > the php root directory is nfs mounted, I 'think' that
> > anywhere php writes a file is local.
> >
> > What would be good is if We could find someone
> > that 'really' doesn't have this problem and see
> > if/how our php.ini or httpd.conf files differ.
> >
> > a new error I got today that I hadn't seen before....
> >
> > ===============================================================
> > An error occured while loading
> > https://mail.math.unm.edu/horde/imp/redirect.php:
> >
> > Found a cyclic link in
> >
>
https://mail.math.unm.edu/horde/imp/login.php?url=/horde/imp/mailbox.php?Horde=52399a6644602fb7df5d4695ce8f698b
> >
> >
> >  _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
> > [_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_X_]
> >
> > Craig Lewis     505 277 6068      help@math.unm.edu
> > UNIVERSITY OF NEW MEXICO MATHEMATICS AND STATISTICS
> > COMPUTER SYSTEMS SUPPORT
> > Humanities Building Office 415
> > Albuquerque, NM 87131
> >
> > On Mon, 12 Aug 2002, Jeremy Freeman wrote:
> >
> >
> >>For some reason, my Apache is randomly segfaulting when I access IMP
> >>3.1. This causes the IE6 to display a "This Page cannot be displayed".
> >>It seems to happen in 1 out of 20 clicks. It doesn't seem to matter
> >>which browser I use.
> >>
> >>Does anyone have any ideas?
> >>
> >>--
> >>Jeremy
> >>
> >>I use a self-signed SSL cert and everything goes over SSL.
> >>
> >>Apache error.log shows:
> >>
> >>[Mon Aug 12 10:33:42 2002] [notice] child pid 4994 exit signal
> >>Segmentation fault (11)
> >>[Mon Aug 12 10:36:23 2002] [notice] child pid 4996 exit signal
> >>Segmentation fault (11)
> >>[Mon Aug 12 10:59:43 2002] [notice] child pid 5511 exit signal
> >>Segmentation fault (11)
> >>[Mon Aug 12 10:59:44 2002] [notice] child pid 5456 exit signal
> >>Segmentation fault (11)
> >>[Mon Aug 12 10:59:48 2002] [notice] child pid 4993 exit signal
> >>Segmentation fault (11)
> >>
> >>I am Running:
> >>
> >>Apache/1.3.26 (Unix) with
> >>PHP/4.2.2
> >>mod_perl/1.27
> >>mod_ssl/2.8.10 OpenSSL/0.9.6e
> >>
> >>PHP compiled with: './configure' '--with-mysql'
> >>'--with-apache=../apache_1.3.26/' '--enable-debug' '--with-gettext'
> >>'--with-imap=../imap-2002.RC2/' '--with-xml'
> >>
> >>All compiled in static.
> >>
> >>Kernel: Linux 2.2.19pre17
> >>
> >>GDB backtrace reveals:
> >>
> >>Program received signal SIGPIPE, Broken pipe.
> >>0x2adb43e4 in write () from /lib/libc.so.6
> >>(gdb) bt
> >>#0  0x2adb43e4 in write () from /lib/libc.so.6
> >>#1  0x2ac2836c in TXT_DB_version () from /usr/lib/libcrypto.so.0.9.6
> >>#2  0x00000000 in ?? ()
> >>(gdb)
> >>
> >>
> >>--
> >>IMP mailing list
> >>Frequently Asked Questions: http://horde.org/faq/
> >>To unsubscribe, mail: imp-unsubscribe@lists.horde.org
> >>
> >
> >
> >
> >
> >
> >
> 
> 
> --
> IMP mailing list
> Frequently Asked Questions: http://horde.org/faq/
> To unsubscribe, mail: imp-unsubscribe@lists.horde.org


--

-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/