Saturday, May 14, 2016

How To Run a Multi-Node Cluster Database with Cassandra ?

Introduction

Apache Cassandra is a highly scalable open source database system, achieving great performance on multi-node setups.
Previously, we went over how to run a single-node Cassandra cluster. In this tutorial, you’ll learn how to install and use Cassandra to run a multi-node cluster on Ubuntu 14.04.

Prerequisites

Because you're about to build a multi-node Cassandra cluster, you must determine how many servers you'd like to have in your cluster and configure each of them. It is recommended, but not required, that they have the same or similar specifications.
To complete this tutorial, you'll need the following:

Step 1 — Deleting Default Data

Servers in a Cassandra cluster are known as nodes. What you have on each server right now is a single-node Cassandra cluster. In this step, we'll set up the nodes to function as a multi-node Cassandra cluster.
All the commands in this and subsequent steps must be repeated on each node in the cluster, so be sure to have as many terminals open as you have nodes in the cluster.
The first command you'll run on each node will stop the Cassandra daemon.
  • sudo service cassandra stop
When that's completed, delete the default dataset.
  • sudo rm -rf /var/lib/cassandra/data/system/*

Step 2 — Configuring the Cluster

Cassandra's configuration file is located in the /etc/cassandra directory. That configuration file,cassandra.yaml, contains many directives and is very well commented. In this step, we'll modify that file to set up the cluster.
Only the following directives need to be modified to set up a multi-node Cassandra cluster:
  • cluster_name: This is the name of your cluster.
  • -seeds: This is a comma-delimited list of the IP address of each node in the cluster.
  • listen_address: This is IP address that other nodes in the cluster will use to connect to this one. It defaults to localhost and needs changed to the IP address of the node.
  • rpc_address: This is the IP address for remote procedure calls. It defaults to localhost. If the server's hostname is properly configured, leave this as is. Otherwise, change to server's IP address or the loopback address (127.0.0.1).
  • endpoint_snitch: Name of the snitch, which is what tells Cassandra about what its network looks like. This defaults to SimpleSnitch, which is used for networks in one datacenter. In our case, we'll change it to GossipingPropertyFileSnitch, which is preferred for production setups.
  • auto_bootstrap: This directive is not in the configuration file, so it has to be added and set tofalse. This makes new nodes automatically use the right data. It is optional if you're adding nodes to an existing cluster, but required when you're initializing a fresh cluster, that is, one with no data.
Open the configuration file for editing using nano or your favorite text editor.
  • sudo nano /etc/cassandra/cassandra.yaml
Search the file for the following directives and modify them as below to match your cluster. Replaceyour_server_ip with the IP address of the server you're currently working on. The - seeds: list should be the same on every server, and will contain each server's IP address separated by commas.
/etc/cassandra/cassandra.yaml
. . .

cluster_name: 'CassandraDOCluster'

. . .

seed_provider:
  - class_name: org.apache.cassandra.locator.SimpleSeedProvider
    parameters:
         - seeds: "your_server_ip,your_server_ip_2,...your_server_ip_n"

. . .

listen_address: your_server_ip

. . .

rpc_address: your_server_ip

. . .

endpoint_snitch: GossipingPropertyFileSnitch

. . .
At the bottom of the file, add in the auto_bootstrap directive by pasting in this line:
/etc/cassandra/cassandra.yaml
auto_bootstrap: false
When you're finished modifying the file, save and close it. Repeat this step for all the servers you want to include in the cluster.

Step 3 — Configuring the Firewall

At this point, the cluster has been configured, but the nodes are not communicating. In this step, we'll configure the firewall to allow Cassandra traffic.
First, restart the Cassandra daemon on each.
  • sudo service cassandra start
If you check the status of the cluster, you'll find that only the local node is listed, because it's not yet able to communicate with the other nodes.
  • sudo nodetool status
Output
Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address      Load       Tokens       Owns    Host ID                               Rack
UN  192.168.1.4  147.48 KB  256          ?       f50799ee-8589-4eb8-a0c8-241cd254e424  rack1

Note: Non-system keyspaces don't have the same replication settings, effective ownership information is meaningless
To allow communication, we'll need to open the following network ports for each node:
  • 7000, which is the TCP port for commands and data.
  • 9042, which is the TCP port for the native transport server. cqlsh, the Cassandra command line utility, will connect to the cluster through this port.
To modify the firewall rules, open the rules file for IPv4.
  • sudo nano /etc/iptables/rules.v4
Copy and paste the following line within the INPUT chain, which will allow traffic on the aforementioned ports. If you're using the rules.v4 file from the firewall tutorial, you can insert the following line just before the # Reject anything that's fallen through to this point comment.
The IP address specified by-s should be the IP address of another node in the cluster. If you have two nodes with IP addresses 111.111.111.111 and 222.222.222.222, the rule on the111.111.111.111 machine should use the IP address 222.222.222.222.
New firewall rule
-A INPUT -p tcp -s your_other_server_ip -m multiport --dports 7000,9042 -m state --state NEW,ESTABLISHED -j ACCEPT
After adding the rule, save and close the file, then restart IPTables.
  • sudo service iptables-persistent restart

Step 4 — Check the Cluster Status

We've now completed all the steps needed to make the nodes into a multi-node cluster. You can verify that they're all communicating by checking their status.
  • sudo nodetool status
Output
Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address      Load       Tokens       Owns    Host ID                               Rack
UN  192.168.1.4  147.48 KB  256          ?       f50799ee-8589-4eb8-a0c8-241cd254e424  rack1
UN  192.168.1.6  139.04 KB  256          ?       54b16af1-ad0a-4288-b34e-cacab39caeec  rack1

Note: Non-system keyspaces don't have the same replication settings, effective ownership information is meaningless
If you can see all the nodes you configured, you've just successfully set up a multi-node Cassandra cluster.
You can also check if you can connect to the cluster using cqlsh, the Cassandra command line client. Note that you can specify the IP address of any node in the cluster for this command.
  • cqlsh your_server_ip 9042
You will see it connect:
Output
Connected to My DO Cluster at 192.168.1.6:9042.
[cqlsh 5.0.1 | Cassandra 2.2.3 | CQL spec 3.3.1 | Native protocol v4]
Use HELP for help.
cqlsh>
Then you can exit the CQL terminal.
  • exit

Conclusion

Congratulations! You now have a multi-node Cassandra cluster running on Ubuntu 14.04. More information about Cassandra is available at the project's website. If you need to troubleshoot the cluster, the first place to look for clues are in the log files, which are located in the /var/log/cassandra directory.

No comments:

Post a Comment