On 7/17/07, Leon Messerschmidt <[EMAIL PROTECTED]> wrote:
1.  The search index is going to be somewhat of a problem.  There will
need to be some extra development to use a central search engine such as
solr.

Yes. Search does not (yet) work in a distributed fashion.


2.  The roller_data directory will need to live on a location (NFS
perhaps) that is accessible to all instances of roller

Correct.


3.  We'll have to turn caching off entirely or look into distributed
caching.

The cache is fully pluggable, so you can plugin a distributed cache.

You can use memcached as a distributed cache via the plugin here:
https://roller.dev.java.net/source/browse/roller/roller_support/tools/memcached/

But we don't have docs for it yet.


4.  We'll have to turn off pinging in all the instances except for one.

Correct.


Anything else that I should be aware of?  Anybody care to help me out?

I think you are mostly on track. Allen had a good post about
distributed Roller a while back, but I was unable to turn it up via
Google.

- Dave

Reply via email to