Apache 1.2.5 & Solaris 2.6 Problems..?

["Ralf S. Engelschall" <rs...@engelschall.com>]

Under my new Solaris 2.6 box I discovered that the 1.2.5 which is running on
this machine always produces some very heavy running processes which slow down
the machine dramatically:

As an example here is the snippet from a "top" display on this machine:

 2322 nobody     1  -5    0 6148K 3268K run   199:12 44.59% httpd
  605 nobody     1  -4    0 5688K 1560K run   734:50 44.10% httpd
                                              ^^^^^^
Actually nearly the same situations occured three times in the last days.
Always two httpd's out of 10 are busy running and the load was increased but
remains equal.  Also the process size remains nearly the same (the 5-6 MB are
typical because I run mod_perl and have a lot of Perl modules pre-loaded).  So
I implicate that these two processes doing some contant never ending loops or
whatever.  Notice the processing time above, too.

The exactly same configuration is running on my FreeBSD boxes quite well for a
long time without such problems. So, perhaps it is related to Solaris 2.6. Has
anybody seen such a situation, too? Any ideas?

                                       Ralf S. Engelschall
                                       rse@engelschall.com
                                       www.engelschall.com

Re: Apache 1.2.5 & Solaris 2.6 Problems..?

[Michael Douglass <mi...@texas.net>]

On Fri, Feb 06, 1998 at 10:39:06AM -0500, Rasmus Lerdorf said:

> >  2322 nobody     1  -5    0 6148K 3268K run   199:12 44.59% httpd
> >   605 nobody     1  -4    0 5688K 1560K run   734:50 44.10% httpd

> Ok, the logical thing to do here is to strace or truss -p these suckers
> and see what the heck they are doing.

Can you provide your Configuration information (egrep -v ^# Configuration)?
I'm interested in seeing the modules you've compiled in as I have a pretty
heavy trafficed web server running on solaris 2.6 and I have yet to see
this happen.  One think I can think of is perhaps someone is using the
.htaccess file pointing a password file to /dev/null and that is apache
going "read-read-read-read-read" in massive loops.  If the process is doing
nothing but reading, output from an lsof -p would be conclusive on what
file it is reading (note: lsof is not preinstalled).

-- 
Michael Douglass
Texas Networking, Inc.

<tnet admin> anyway, I'm off, perl code is making me [a] crosseyed toady

Re: Apache 1.2.5 & Solaris 2.6 Problems..?

[Dirk-Willem van Gulik <di...@jrc.it>]

On Fri, 6 Feb 1998 rse@engelschall.com wrote:

I saw something similar recently; using gdb and breaking I found
the cild in a loop within the perl interpreter (which was linked
against various older db-dbm libs.). Recompiling the later solved
it (DB_Any &al).

Dw
 
> Under my new Solaris 2.6 box I discovered that the 1.2.5 which is running on
> this machine always produces some very heavy running processes which slow down
> the machine dramatically:
> 
> As an example here is the snippet from a "top" display on this machine:
> 
>  2322 nobody     1  -5    0 6148K 3268K run   199:12 44.59% httpd
>   605 nobody     1  -4    0 5688K 1560K run   734:50 44.10% httpd
>                                               ^^^^^^
> Actually nearly the same situations occured three times in the last days.
> Always two httpd's out of 10 are busy running and the load was increased but
> remains equal.  Also the process size remains nearly the same (the 5-6 MB are
> typical because I run mod_perl and have a lot of Perl modules pre-loaded).  So
> I implicate that these two processes doing some contant never ending loops or
> whatever.  Notice the processing time above, too.
> 
> The exactly same configuration is running on my FreeBSD boxes quite well for a
> long time without such problems. So, perhaps it is related to Solaris 2.6. Has
> anybody seen such a situation, too? Any ideas?
> 
>                                        Ralf S. Engelschall
>                                        rse@engelschall.com
>                                        www.engelschall.com
> 

Re: Apache 1.2.5 & Solaris 2.6 Problems..?

[Rasmus Lerdorf <ra...@lerdorf.on.ca>]

> Under my new Solaris 2.6 box I discovered that the 1.2.5 which is running on
> this machine always produces some very heavy running processes which slow down
> the machine dramatically:
> 
> As an example here is the snippet from a "top" display on this machine:
> 
>  2322 nobody     1  -5    0 6148K 3268K run   199:12 44.59% httpd
>   605 nobody     1  -4    0 5688K 1560K run   734:50 44.10% httpd

Ok, the logical thing to do here is to strace or truss -p these suckers
and see what the heck they are doing.

-Rasmus