Hi Andre,

I want to preface this with the fact that I haven't clustered JS2, nor
JS1.  In fact, I don't even use JS2.

That said, I wanted to comment on how I'd design this system, or at
least start.

--- Andre Bonhote <[EMAIL PROTECTED]> wrote:

> Hi!
> 
> It's my first post, and I am completely new to Jetspeed, so bear with
> me
> please.
> 
> The idea is to deploy a portal on n machines, where n is 4 at the
> moment. We would like to have some pretty load balancers in front of
> them, the boxes are located in two countries. As this is usual for
> good
> portals, the user/customer should not care about where he is.
> 
> The jetspeed-2 installation will access an oracle 10 database. Now
> there's my question: Since it doesn't make sense (IMHO) to put the
> oracle beast on all the 4 machines, is it possible to have all 4
> jetspeed installations access the same database? Or even, the same
> tablespace? Or do I have to create users for each server?

You shouldn't have to create database users for each server.  You
should be able to set up oracle to live on a fifth box, and have each
JS server connect over tcp/ip.  Set up your tnsnames.ora on each box
and it should be easy.

> Is there a clustering guide somewhere? I am quite stuck at the
> moment,
> to be honest.

If I were to want to cluster JS2, I'd first make sure tomcat was
sharing session information across the cluster:
http://jakarta.apache.org/tomcat/tomcat-5.0-doc/cluster-howto.html
(this message will provide some comfort, but I'm not sure if it applies
to JS2
http://nagoya.apache.org/eyebrowse/[EMAIL PROTECTED]&msgId=500554)

Then I'd disable the data cache of JS2, if there is any.  You can test
that easily enough if you get two boxes running with a simple portlet
that changes a row in the db--see if the other change is reflected.

Now, this will of course hurt performance, since you'll be accessing
data across the network every time, but it's a cheap way to cluster. 
I'm really not sure what the JS2 data layer looks like.  Checkt that'
it looks like it might be Torque from the docs on the site.  In that
case, you'll probably be interested in this document:
http://db.apache.org/torque/managers-cache.html
which might help you use torques cache in a clustered environment.
(Not sure what version of torque is being used, though.) 

Also, I'd recommend you develop, or at the very least test, with at
least two machines, as having a dev environment similar to production
can save you a lot of grief.

Good luck and let us know how it goes.

Dan

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to