From: Suman Nath (sknath@cs.cmu.edu)
Date: 02/04/03
What I had in my mind is, the "install" script will compile and install
the code on a machine. The install script will be run just once for a
machine. In contrast, the setup script will be run whenever we want to run
the PT service. It should assume that the "install" script has already
been executed on the partitipating machines, and hence start the SA, OA,
backened CannedQuerries, xindice etc. in all the participating machines
and populate them with the schema. The goal of using such "setup" script
to run the whole service from one console (instead of doing ssh to each of
the machines, and start OA, SA etc. on them) when necessary.
On Tue, 4 Feb 2003, Mahim Mishra wrote:
>
> I don't clearly understand how the "install" and "setup" scripts are to be
> organized. Reading the project description makes it seem that we have to
> run the "install" script on both machines, and the "setup" script on one
> machine. I think that the following structure seems the most logical then:
>
> 1. Let the install script compile and copy the files, and start up the sa,
> oa and xindice
>
> 2. Let the setup script populate the OAs with the schema, subscribe the
> sensletes, and start up the front-end.
>
> This is stretching the purpose of the install script a little, but the
> alternative (getting the setup script, running on one machine, to start up
> the irisnet infrastructure on the other) seems far more painful to me. Any
> comments?
>
> Mahim
>
>
> --
> Mahim Mishra
> Graduate Student, CS Deptt, 8203 Wean Hall, CMU
> School of Computer Science, ph :+1-412-268-3562
> Carnegie Mellon University. www :http://www.cs.cmu.edu/~mahim
>
>
This archive was generated by hypermail 2.1.5 : 06/10/04 EDT