If you are not working in or running out of the feature/cassandra-backend branch feel free
to ignore this as the following info is specific to the feature/cassandra-backend branch
right now.
There has been a good bit of infrastructure put in place for deploying a Cassandra cluster
for automated tests and for the dev-container. When you build and deploy the
dev-container, a cluster will automatically be deployed and started for you. There is
nothing extra you have to do to get a functioning dev-container. That deployment code that
is currently used was implemented well before rhqctl. The rhqctl script does not yet have
support for dev-container deployments. I plan to update rhqctl so that it can be used for
both prod and dev deployments. Then we have a common set of tools across deployments. The
other benefit of using rhqctl is that it offers a lot more functionality than what is
currently used for dev-container deployments.
We have been using out of the box heap settings for Cassandra. These are determined by the
cassandra-env.sh script which is part of the Cassandra distro. There is no equivalent
script for Windows so I'm not sure defaults are used with Windows. The
cassandra-env.sh script will try to use 1/4 of your RAM for the heap; so, if you have 8
GB, it will give Cassandra 2 GB. For dev-container deployments and certainly for automated
tests, we can get by with much smaller heaps.
I pushed some changes that lower the dev-container defaults and also make them
configurable. The default max heap for the dev-container cluster nodes is now 512 MB. You
can change this by editing rhq-server.properties before starting your dev-container for
the first time after it has been (re)built. Uncomment and set the
rhq.cassandra.max.heap.size and rhq.cassandra.heap.new.size properties. If you want to
change the heap settings after the initial deployment, then you will need to edit
<RHQ_HOME>/cassandra/node{0,1}/conf/cassandra-env.sh.
- John
Show replies by date