Mailing-List: contact dev-help@httpd.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@httpd.apache.org
Sender: gregames@Mail.MeepZor.Com
Message-ID: <3C337BBA.9AF3678F@remulak.net>
Date: Wed, 02 Jan 2002 16:29:30 -0500
From: Greg Ames <gregames@remulak.net>
MIME-Version: 1.0
To: dev@httpd.apache.org
Subject: Re: 2.0.30-dev load spiking [was: upgrade to FreeBSD 4.5-PRERELEASE]
References: <3C336A1D.5537B116@remulak.net> <3C3371B6.4070608@cnet.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Brian Pane wrote:

> Do you have a way to take a snapshot of each httpd process's stack
> backtrace?  On Solaris, I'd do this by running /usr/proc/bin/pstack
> on each pid; I don't know if FreeBSD has a similar functionality.

ummm, let's see:

[gregames@daedalus gregames]$ pstack
bash: pstack: command not found
[gregames@daedalus gregames]$ ls /usr/proc
ls: /usr/proc: No such file or directory

nope, sorry.  Of course we could get backtraces with gcore, but would
probably fill up the filesystem in the process.  

Actually, if we had a way of dumping just the two processes that own the
two CPUs when we get one of these spikes, it might be feasible. 
Especially if the trap/shell script/whatever would exit once it did its
thing, and didn't dump itself.

> This would give us a picture of what all those runnable processes
> are doing.

The processes that aren't dispatched are probably innocent bystanders,
at least until they get ahold of a CPU.  But I suppose some of them
could have been doing bad things in previous time slices.

Greg