Re: Release Schedule

From: Robert Collins <robert.collins@dont-contact.us>
Date: 22 Aug 2001 07:53:40 +1000

On 21 Aug 2001 13:33:39 -0500, Joe Cooper wrote:
> For you, Rob, I'll make some time. ;-)

Thank you.
 
> Seriously, though. I'm always extraordinarily busy, but I've got a
> bench set up right now anyway. As long as it doesn't cause any problems
> that need to be tracked down, I can start some runs and see how they go.

I understand - time is a serious problem for me too. Unfortunately I
don't have the resources at home - where I hack - to do polygraphing :[.
So this is much appreciated.

> What branch should I grab for this? Any configuration details I should
> know about?

On sourceforge, the branch external_logger is the one. It does add a new
config option for the log program to be used, but the defaults will work
fine. What it does is log everything through an external helper - logd
by default - which is allowed to do anything it wants, as long as it
keeps the pipe nice and empty by read()ing lots.

This is needed to allow rotates and reconfiugres with ntlm - ntlm
helpers cannot be shut down synchronously, which introduces a lovely
little race with the debug log file, preventing rotate and reconfigure
completely on win32, and causing log entries into the old log file on
unix.

What I'm interested in is the behaviour at high log entry rates -
access.log uses a helper as well (to allow the log-to-sql that has come
up before) so simply pushing the squid, seeing if it goes as high as
you'd have hoped :] and then seeing if access.log did catch every
request should a good trial...

Thanks again!
Rob

> --
> Joe Cooper <joe@swelltech.com>
> Affordable Web Caching Proxy Appliances
> http://www.swelltech.com
>
Received on Tue Aug 21 2001 - 15:53:28 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:14 MST