Moving data between different Cassandras if you're running them using Docker

backup_cassandra.py

#!/usr/bin/python3
import os
import sys

"""
A tool to copy a named snapshot of a cassandra to a target rsynced thing.
Will use ssh to create target directories

Requires: rsync ssh

Keyspaces with names starting with system will be omitted.

Usage:

- nodetool snapshot -t name-of-the-snapshot
- backup_cassandra.py target_host:/base/dir/on_the_target_host /local/path/to/cassandra/data name-of-the-snapshot
- backup_cassandra.py target_host:/base/dir/on_the_target_host /local/path/to/cassandra/data name-of-the-snapshot

Additionally if you define an environment variable called EXCLUDE_KEYSPACES keyspaces mentioned won't be synced
Additionally if you define an environment variable called EXCLUDE_TABLES keyspaces mentioned won't be synced

"""

if __name__ == '__main__':
    local_cassandra_directory = sys.argv[2]
    keyspaces = os.listdir(local_cassandra_directory)
    target_host, target_host_dir = sys.argv[1].split(':')
    snap_name = sys.argv[3]
    if 'EXCLUDE_KEYSPACES' in os.environ:
        for keyspace in os.environ['EXCLUDE_KEYSPACES'].split(' '):
            keyspaces.remove(keyspace)
    keyspaces = [ks for ks in keyspaces if not ks.startswith('system')]

    tables_to_kill = os.environ.get('EXCLUDE_TABLES', '').split(' ')

    for keyspace in keyspaces:
        for table_directory in os.listdir(os.path.join(sys.argv[2], keyspace)):
            table_name = '-'.join(table_directory.split('-')[:-1])
            if table_name in tables_to_kill:
                continue
            print('Proceeding with {keyspace}:{table_name}'.format(**locals()))
            os.system("ssh {target_host} mkdir -p {target_host_dir}/{keyspace}/{table_name}".format(**locals()))
            src_path = os.path.join(local_cassandra_directory, keyspace, table_directory, 'snapshots', snap_name)
            os.system('rsync --progress -r -v {src_path}/* {target_host}:{target_host_dir}/{keyspace}/{table_name}'.format(**locals()))
  

restore_cassandra.py

#!/usr/bin/python3
import os
import sys
import subprocess

"""
A tool to restore the data to Cassandra

Requires: docker + all relevant schemas to be available

Usage:

- restore_cassandra.py cassandra-container-name RPC_ADDRESS /var/lib/cassandra/local/container/site/of/backup/made/by/backup_cassandra.py
"""
def do_exec(path, **kwargs):
   path = path.format(**kwargs)
   cmd = 'docker exec -it {container_name} '+path
   cmd = cmd.format(**kwargs)
   print('Attempting', cmd)
   a = subprocess.check_output(cmd, shell=True)
   a = a.decode('utf-8').replace('\t', ' ').replace('\r', ' ').replace('\n', ' ')
   return [k for k in a.split(' ') if k.strip()]


if __name__ == '__main__':

  container_name = sys.argv[1]
  cassandra_host = sys.argv[2]
  backup_path = sys.argv[3]

  keyspaces = do_exec('ls {backup_path}', **locals())

  for keyspace in keyspaces:
    tables = do_exec('ls {backup_path}/{keyspace}', **locals())
    for table in tables:
      do_exec('sstableloader -d {cassandra_host} {backup_path}/{keyspace}/{table}', **locals())