Restore Cassandra data

Owned by Rajesh Rajendran
Nov 25, 2021
2 min read

Scenario: We're taking backup ( snapshot of data from each node of the cluster with naming node-1-data-DD-MM-YY and uploading it to storage( in our case Azure Blob Storage). The following steps depict how to restore the data.

Note: This works only on a cluster with the same number of nodes, which means if you have a 7 node cluster; To restore the data you need a 7 node cluster. Though you don't need the same size VM. For example, if you have a 16core 64G machine in prod for a single node, you need only 2 core 4G machines to restore the data. But the node count should be the same.

Note: For the procedure to be clear, I am assuming that you are going to restore a 7 node cluster backup. Cassandra directory is /var/lib/cassandra/ and Cassandra configuration file is /etc/cassandra/cassandra.yaml

Steps:

  1. Download the data for each node respectively. Means node-1 data into the new node-1 machine and node-2 data in the new node-2, so on and so forth.

Follow the steps in all nodes

  1. Stop the Cassandra cluster: sudo systemctl stop cassandra

  2. Remove all data: sudo rm -rf /var/lib/cassandra/*

  3. In every backup folder, there will a tokenring.txt file which contains the token ring for that node. Copy the content of that file and paste in /etc/cassandra/cassandra.yaml as initial_token: <data copied>.
    Ref initial_token: https://docs.datastax.com/en/cassandra-oss/2.1/cassandra/configuration/configCassandra_yaml_r.html#configCassandra_yaml_r__initial_token

  4. Change the permission of the cassandra directory
    sudo chown -R cassandra:cassandra /var/lib/cassandra

  5. Start Cassandra: sudo systemctl start cassandra
    After some time, check nodetool status and all nodes should be UN status

Run in only one node

  1. Restoring schema: In any node, go to backupfolder/cassandra_backup/ there will be a db_schema.sql file.
    To restore the schema: cqlsh -f db_schema.sql. Once that operation is done, follow the below steps.

Run in all nodes

  1. Stop Cassandra: sudo systemctl stop cassandra

  2. Download the restoration script

  3. Restore the data: sudo python3 ./cassandra_restore_v2.py --snapshotdir /path/to/backup_dir/cassandra_backup --datadirectory /var/lib/cassandra/data
    It'll run a bunch of operations, and copy the data into the Cassandra data folder.
    Note: These copy operations are hard links, so you don't need double the space of data.

  4. Check the data size
    du -sh /var/lib/cassandra/data

  5. Change the permission of data: sudo chown -R cassandra:cassandra /var/lib/cassandra/*

  6. Start Cassandra sudo systemctl start cassandra

Note: if any of the Cassandra nodes didn't start do a stop and start Cassandra. This is because all the nodes started at the same time.