Cassandra Installation Questions in AWS

Question

We are planning to install Cassandra in AWS EC2 instance. Single Node cluster was installed. If we need to install multi node cluster, please help us on the below questions.

1. During default single node install, we logged into the AWS instance using Public DNS through Putty and installed the Cassandra through CLI. How to add new nodes for Cassandra in AWS instance for making it as a multi node cluster?

2. How to get/determine the IP addresses of additional Cassandra nodes? Does AWS has any features to find this? For example, AWS Console or CLI commands?

3. In every blog, it says copy the same YAML file to other nodes. How to login to each node separately and do this?

4. Are there any differences installing Cassandra in EMR or EC2?

Answer 1

I have done this on EC2 (can't tell you anything about EMR), so I can definitely help you from that perspective. We used our own, custom tool to provision the AWS nodes, and then deployed with Chef.

The nice part about Chef, was that your yaml and other configs would then get consistently deployed to each node in the cluster. We would then have a "wrapper cookbook" around our base cookbook (base cookbook installed Cassandra, disabled swap, and set a few other OS-dependent settings), where we would put all of our settings. The wrapper cookbook would also retrieve the IP addresses of the node, and use it to set the appropriate fields in the configs.

The tricky part was the seed logic. We would always build the first node, and then wait long enough for it to be assigned an IP. The Chef cookbook had logic to add the current IP to the seed list if it was empty. Once we had the initial seed IP, we could then make a change to our wrapper cookbook to use that as a seed list.

You'll definitely want to use some kind of deployment tool for this. Installing via CLI on each is the hard way, and is prone to errors. Hope that helps.