Ive come up with an elegent solution, i have setigate running on the server
machine and I run 2 instances of seti using shellexecute. My only problem
now is caching enough units on my 56k!
Phil.


"John Clark" <fredclark@consltec.demon.co.uk> wrote in message
news:+CwBSLBRuzV$Ewal@consltec.demon.co.uk...

In article <gq7elvgmmi06jicc338ok4oq41fimm5j0u@4ax.com>, f/fgeorge
<an89765478@anon.penet.de> writes

On Wed, 03 Sep 2003 15:34:21 GMT, "Stratcat�" <nobody@no.org> wrote:

I solved this by installing one instance of SetiWatch in the folder with
everything else. I now use SetiWatch as my results log. I saw a post by
f/fgeorge suggesting SetiWatch as well, although I'm not sure how he
configures his. Either way, this was my solution, FWIW.

I do it the same way you do. I have one instance of Seti Driver, and
one instance of SetiWatch. I do have one minor change though....I have
4 selected in the "desired cache size" section of SetiDriver. It was
suggested by Mike Ober the writer of the program. Rarely both work
units will finish at exactly the same time, or at least within the
time it takes to transmit and get a new work unit. When that happens
the program will hang if there are no work units in the cache. Having
a local cache such as SetiQue will not help. So he suggested that on
machines with more than 1 cpu or with HT enabled that the cache size
in SetiDriver be double what is actually being processed.
IOW, in your case it would be 4. Mine is also 4 since I have an actual
dual processor machine.

Following these suggestions is the way I play it. However, I have
changed the situation slightly. My small network consists of -

1.      A single proc P3 750, o/c to 840. This has no cash over the WU
        being processed. This runs a single instance of the CLI, with
SETI Driver and SETI Spy.
2.      I have a dual proc P3 933@933, and 2 instances of the CLI and
SETI Driver/Spy.
3.      Finally, I have a dual proc Xeon @ 2.8 Ghz with HT enabled. This
        runs 4 instances of the CLI, with Driver and Spy.

The whole system is tied together with SETI Queue, as my proxy server
and stats generator.

NOTE: SETI Watch would also do the job OK.

In each of the versions of SETI Driver I have the Proxy server box
ticked, and the TCP/IP address: port 5517 enabled.

The system works fine, collecting all the WUs as these are finished and
dishing out the new WUs. None of the SETI Driver cashes are above the
number needed to process the appropriate number of units for each of the
network PCs.

In the evening SETI Queue sends the day's collection of completed WU
results off and automatically collects the replacements.

This set up may or may not suit you?
-- 
John Clark