httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From (Rasmus Lerdorf)
Subject [PHP-DEV] Bug #693: Apache process runs out of control (fwd)
Date Tue, 25 Aug 1998 13:52:50 GMT
This bug report was just submitted to the PHP3 bug database.  Looking
through the Apache sources, signal(SIGCHLD, SIG_DFL) is sometimes called
in order to fix an HP_UX bug according to the comments, yet there is no
#if HP_UX around the call.  Anybody know the full story here?

---------- Forwarded message ----------
Date: 25 Aug 1998 13:44:34 -0000
Subject: [PHP-DEV] Bug #693: Apache process runs out of control

Operating system: Solaris 2.6.1
PHP version:      3.0.3
PHP Bug Type:     Oracle related
Bug description:

It seems that after I access a page with using oracle
access several times, the next time I run a cgi script,
the server process goes out of control, using all the
cpu.  (See bugs 375 and 450).

I think it is Oracle related.  It only happens after I have
accessed a page which talked to our oracle database.

I finally had some time to look into this last week, and I
have a hack which seems to have fixed the problem.

Doing a truss on the process, it seems that there are lots
of SIGCHLDs going to it.  (I can give send the output of
this if you like.)  So, in the file main.c, at the ttop of the
function php3_parse (about line 1427), I added the following line:

signal (SIGCHLD, SIG_DFL);

I'm not much of a unix programmer, but I have played
around with it before.  After adding the line above, it
fixed my problems.

I don't know where the problem really is, though.  Is it
a problem with php?  apache?  the oracle libraries?

Will my fix above break anything else? (it hasn't so far).

I'm running apache 1.3.1 and php 3.0.3.

My configure script for php is below:



PHP Development Mailing List
To unsubscribe send an empty message to
For help: 

View raw message