On 4/20/05, Henrik Nordstrom <[EMAIL PROTECTED]> wrote:
> > The rest are url_regex which involve urls or ports (for the CONNECT)
> > later defined in the http_access rules. So essentially we have a
> > population of users and we want to restrict what they can access
> > depending on what group they're in. Group membership is determined by
> > ldap lookups.
> 
> Why url_regex?
> 
> For CONNECT there is exacly zero reasons to use url_regex.
> 
> In terms of CPU usage url_regex is several orders of magnitude heavier
> than the other acl types.

I'll take a look at it. I think it was to keep the format simpler
because the majority of the acls are url_regexes. But I don't think
this is the problem.

> 
> > Never any problems with CPU usage, these are like dual p3 1.3 Ghz, so
> > it's more than enough muscle.
> 
> Squid only runs on one CPU. What means that 50% CPU usage reported on your
> system is 100% CPU usage by Squid..

It's essentially 1% of 1 CPU usage.

I haven't looked at how the ACLs are evaluated and if re-ordering them
would help. Is this a worthwhile idea?

Thien

Reply via email to