Quoting Justin Hopkins <[email protected]>:
I've been using that image on EC2 recently, so I thought I'd add
what I've learned so far:
* A 'small' instance is just barely enough to run it. It's *very* slow.
* opensrf services do not start up well. I need to log in each time
I start the instance and run osrf_ctl.sh -l -a stop_all/
osrf_ctl.sh -l -a start_all
There is no automated startup script for OpenSRF/Evergreen, i.e.
something that lives in /etc/init.d/ and so on, so that services can
be started and stopped automatically at system boot and shutdown. Part
of the reason for this is that you may want to run different services
on different servers in a mulit-server environment. That said, if
someone wants to create such scripts for Ubuntu, Debian, or Fedora
Core, then patches are most welcome.
To answer the OP's question, I use a virtual machine on an extra
server in the office for my main development platform. Others use it
to test and demo new features. It has access to 6GB of RAM and 80GB
hard disk space. Search can be a bit slow at times, but other than
that it works really well.
To demo Evergreen, you don't need a super powerful machine, depending
on your needs. I also have an old desktop with 2GB of RAM that I am
using to work on porting Evergreen to FreeBSD. It runs rather well
with the concerto data set, but again search can be slow.
So, if you just want to demonstrate that Evergreen works and what
features it has, you can get away with an old workstation. If you want
to demonstrate that it has some speed, I'd recommend a low end server
with 8GB or more of RAM.
If you set up a virtual machine in the cloud or on a pre-existing
server, I'd recommend giving it at least 8GB of RAM and 80GB of disk
space. You won't need that much disk space if you load the concerto
data set, but depending on how you use the demo machine, disk space
can disappear in a hurry.
HtH,
Jason
--
Jason Stephenson
Assistant Director for Technology Services
Merrimack Valley Library Consortium
Chief Bug Wrangler, Evergreen ILS
