This page provides instructions to set up a Teamwork Cloud (TWCloud) Cluster on your system. A TWCloud Cluster is composed of two clustering layers, TWCloud and Cassandra layers.
To set up a TWCloud Cluster, you need to do two separate tasks
An illustration of TWCloud Cluster and Cassandra Cluster nodes within TWCloud Cluster.
Setting up a Cassandra Cluster
Prior to establishing a Cassandra cluster, you need to determine the following.
- The initial number of nodes in the cluster.
- The IP address of each node.
- Determine which node to be the seed.
When setting up a Cassandra node, you need to configure a seed that indicates an initial contact point for the new node. So, you can configure the seed using the IP of any existing active node in the cluster. Use the node's own IP when configuring the first node. For example, if you are configuring a three-node cluster and the nodes' IPs are 10.1.1.101, 10.1.1.102, and 10.1.1.103. Select one, for example 10.1.1.101, as the seed. While you are configuring cassandra.yaml, specify 10.1.1.101 as the seed value for all of the three nodes.
Now that you understand what a seed is and how to configure it, follow the instructions to install and configure Cassandra on Windows or Linux Operating System.
Please note that listen_address and broadcast_rpc_address are still their machine.
| Machine IP | 10.1.1.101 | 10.1.1.102 | 10.1.1.103 |
|---|---|---|---|
| seeds | 10.1.1.101 | 10.1.1.101 | 10.1.1.101 |
| listen_address | 10.1.1.101 | 10.1.1.102 | 10.1.1.103 |
| broadcast_rpc_address | 10.1.1.101 | 10.1.1.102 | 10.1.1.103 |
After installing and configuring all nodes, start the seed node. Verify that it is operational by issuing the command "nodetool status".
Once the seed node is operational, you can proceed to start the other nodes in the cluster. You must allow the node to join the cluster before proceeding to the next node. After all nodes have started, the "nodetool status" command will display results as shown below.
Datacenter: datacenter1=======================Status=Up/Down|/ State=Normal/Leaving/Joining/Moving-- Address Load Tokens Owns (effective) Host ID RackUN 10.1.1.101 6.7 GB 256 35.1% b33c603a-95c7-426d-9f7b-ebad2375086a rack1UN 10.1.1.102 6.34 GB 256 32.5% 16f40503-4a65-45fe-9ee7-5d942506aa87 rack1UN 10.1.1.103 6.15 GB 256 32.4% 2d90c119-08a4-4799-ac73-0440215d0b18 rack1
Node Consistency
In order to ensure that all nodes are consistent, please run the "nodetool repair" command nightly. This can be scheduled as a crontab job or via Windows Task Scheduler. This only needs to be executed on a single node.
Setting up a TWCloud Cluster
Before setting up a TWCloud Cluster, you need to determine the following.
- The initial number of nodes in the cluster.
- The IP address of each TWC node.
- Prepare a list of IPs of Cassandra nodes.
The instructions below use the following sample environment for ease of understanding.
- 3 nodes TWC cluster
- TWC node IP address are 10.1.1.111, 10.1.1.112, and 10.1.1.113
- All nodes will be used as seed nodes
- IP address of Cassandra nodes are 10.1.1.101, 10.1.1.102, 10.1.1.103
Setting up the TWCloud Cluster involves 2 groups of parameters in application.conf.
- TWCloud clustering parameter
- akka.cluster.seed-nodes
This parameter indicates the initial contact points for the cluster.
If you install TWCloud using the installer file, you will be asked to provide the seed node IP during installation process and the value will be configured in application.conf.
If you install using the zip file, you will need to manually configure the parameter in application.conf. Search for the following:
seed-nodes = ["akka.tcp://twcloud@${seed-node.ip}:2552"]
Replace ${seed-node.ip} with the IP addresses of the seed nodes, so it should look similar to the following:
seed-nodes = ["akka.tcp://twcloud@10.1.1.111:2552","akka.tcp://twcloud@10.1.1.112:2552","akka.tcp://twcloud@10.1.1.113:2552"]
- Cassandra configuration parameters
- esi.persistence.cassandra.connection.seeds
The value of this parameter is a list of Cassandra node IP addresses. You can find this parameter in application.conf by looking for the following.
# List of comma delimited host.
# Setting the value as localhost will be resolved from InetAddress.getLocalHost().getHostAddress()
# Ex. seeds = ["10.1.1.123", "10.1.1.124", "10.1.1.125"]
seeds = ["localhost"]
As you can see, the default value is [“localhost”], which is only suitable for a single node server where both TWCloud and Cassandra are deployed on the same machine. According to our sample environment, you should change it to the following.
seeds = ["10.1.1.101", "10.1.1.102", "10.1.1.103"]
- esi.persistence.cassandra.keyspace.replication-factor
This parameter defines the Cassandra replication factor for the “esi” keyspace used by TWCloud. The replication factor describes how many copies of your data will be written by Cassandra. For example, replication factor 2 means your data will be written to 2 nodes.
For a three-node cluster, if you would like the cluster to be able to survive 1 node loss, you will need to set the replication factor to 3.
persistence {cassandra {keyspace {replication-factor = 3}
Please note that this configuration will be used only for the first time TWCloud connects to Cassandra and creates a new esi keyspace. Changing the replication factor after the keyspace has already been created is rather a complex task. Read this document if you need to change it.
Note
TWCloud uses QUORUM for both write and read consistency level.
Click here for a detailed explanation about data consistency.
To start up the TWCloud cluster, first start the server on the seed machine and wait until you see a message similar to the following in the server.log.
INFO 2017-02-15 10:57:08.409 TWCloud Cluster with 1 node(s) : [10.1.1.111] [com.nomagic.esi.server.core.actor.ClusterHealthActor, twcloud-esi.actor.other-dispatcher-31]
Then you can start the server on the remaining machines. You should see the following messages in the server.log which shows all 3 nodes are forming the cluster.
INFO 2017-02-15 10:58:23.956 TWCloud Cluster with 2 node(s) : [10.1.1.111, 10.1.1.112] [com.nomagic.esi.server.core.actor.ClusterHealthActor, twcloud-esi.actor.other-dispatcher-18]INFO 2017-02-15 10:58:25.963 TWCloud Cluster with 3 node(s) : [10.1.1.111, 10.1.1.112, 10.1.1.113] [com.nomagic.esi.server.core.actor .ClusterHealthActor, twcloud-esi.actor.other-dispatcher-18]