httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ralf S. Engelschall" <...@engelschall.com>
Subject Re: mod_setclass.c (was Re: gcc 2.8.1 not to blame)
Date Wed, 22 Apr 1998 18:57:53 GMT

In article <353E3630.AF91470@algroup.co.uk> you wrote:
> Rob Hartill wrote:
>> 
>> On Tue, 21 Apr 1998, Brian Behlendorf wrote:
>> 
>> > I came in this morning to find another 4 apache processes running away with
>> > CPU.
>> >
>> >   PID USERNAME PRI NICE SIZE    RES STATE    TIME   WCPU    CPU COMMAND
>> > 29266 nobody   91   0   828K   904K RUN     50:49 20.87% 20.87% apache
>> > 29248 nobody   91   0   876K   884K RUN     50:49 20.87% 20.87% apache
>> > 29149 nobody   92   0   852K   904K RUN     50:44 20.56% 20.56% apache
>> > 29267 nobody   92   0   828K   920K RUN     50:54 20.52% 20.52% apache
>> >
>> > ktrace returned nothing.  sending various signals via 'kill' to force a
>> > core dump produced nothing.  This is the CVS tree as of yesterday.  The
>> > /status output coresponding to those 4 processes looks like:
>> 
>> While it's much better to find the cause and fix it, stopping the
>> runaways overnight might be desirable in the short term.
>> 
>> Here's my FreeBSD module that puts Apache under the restrictions of
>> a /etc/login.conf class.  Give it some limits for CPU time and memory
>> and the nasty runaways will have the legs kicked from under them..

> I like this. We should have it in the distribution.

Yes, its a useful and tiny module. We could add it as modules/experimental/
to make sure mod_mmap_static doesn't feel alone is this dir ;-)
Without joke: I would appreciate this in the distribution, too. +1

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

Mime
View raw message