Re: More IMS

On Wed, 21 Apr 2004, Leeann BENT wrote:

> I'm not sure I understand. The new lifetime of the object should be:
> ((newdate - last-modified)*pct), assuming the last-modified heuristic is
> used? Is this where the new date is used in the freshness calculations?

Yes.

> Does this mean that the object is fresh from now for the calculated ttl or
> from when it first entered the cache for the caclculated ttl.

From now for the calculated ttl.

On a successful cache validation all of

   timestamp (date object last seen fresh, measured in local clock)
   lastmod (last-modified header)
   expires (expires header)

is updated.

The actual object headers is not updated however (Bug #14 I think..), but
all parameters used for freshness calculations is.

> > Should not be the case.
>
> It appears that this is what I'm seeing in the latest release. I'll check
> a previous release to see if this is the case. Once an object is expired,
> Squid3 is always emitting the IMS request, but I'm not sure why.

Maybe the call to storeTimestampsSet on the original object while
processing the 304 has gone missing, or the mem objects reply structure is
not properly updated before..

> They show two calls to httpSendRequest. I've attached the log.

Ok,

> > Is this two requests initiated by Squid, or two requests initiated by the
> > client?
>
> There is no client. I'm telneting in and requesting the object this way:
>
> % telnet localhost 3128
> GET http://www.myserver.com/test.txt HTTP/1.1
> <cr>
> <cr>

Well. there is a client (you).. but this answers my question. There is
only a single request initiated by the client.

Regards
Henrik
Received on Wed Apr 21 2004 - 15:08:32 MDT