Felix Karpfen <[EMAIL PROTECTED]> wrote:
> I have recently switched my business to a different ISP who offers (on
> paper) better value for money.
>
> However, since the switch, I have repeatedly `timed-out' my connection
> to the ISP while WWWOFFLE was fetching its `outgoing' and `monitored'
> pages.
>
> It is on the cards that WWWOFFLE stalled on several attempted fetches
> (because of poor lines) and this stall was interpreted by the ISP server
> as a `timeout'. Attached to this email is an editied log of one crash.
>
> My question is:
>
> Which (if any) entries in wwwoffle.conf are worth modifying to
> <avoid/minimise> such timeouts?
The main timeout is the socket timeout. Typically this will be two
minutes. This means that the worst case time that you will be
connected to your ISP without any packets travelling is 2 minutes.
Obviously if you have more than one WWWOFFLE fetch process running at
once then the chances are that this 2 minutes time will be reduced
proportionately.
The other timeouts are the DNS timeout (usually 60 seconds) and the
connect timeout (usually 30 seconds). Since both of these are smaller
than the default socket timeout they will not be important.
The best thing that you can do is to find out what the ISP timeout is
and set the socket timeouts to be less than this.
> I am concerned not merely with the default entries that relate
> specifically to `time', but also to the entries listed under
> `FetchOptions' and ModifyHTML (eg. images, scripts, blink etc).
These options would have no effect on the timeouts that you are
seeing.
> Also, related to the above, I presume that the FetchOptions entries
> refer specifically to WWWOFFLE's response to the `fetch' command.
> Do entries under this heading also impact on what is fetched at the
> direction of the browser (via the localhost proxy) while online?
No, the FetchOptions are just those items referenced in the HTML of
the downloaded page (caused by a 'wwwoffle -fetch') that are also
fetched.
The other method of reducing timeouts is to run some other program
that generates network packets and keeps the ISP connection alive.
This would obviously be unethical if the ISP timeout is designed to
stop people from keeping connections alive permanently and the timeout
is a reasonable length of time that does not cause extra expense
making multiple connections.
--
Andrew.
----------------------------------------------------------------------
Andrew M. Bishop [EMAIL PROTECTED]
http://www.gedanken.demon.co.uk/
WWWOFFLE users page:
http://www.gedanken.demon.co.uk/wwwoffle/version-2.6/user.html
