You are not reading the most current version of the documentation. If you want up-to-date information, please have a look at 2.5 .

Extract schema from system_schema keyspaceΒΆ


  1. You have the snapshot tag.

  2. The snapshot contains system_schema keyspace.


  1. List files of snapshot we want to restore (replace <node_id> with ID of one of the nodes in the cluster.

    sctool backup files --cluster my-cluster -T sm_20200513104924UTC -K system_schema --with-version | grep "<node_id>" > backup_files.out
  2. Create a directory, and recreate table directories in it.

    mkdir scylla
    cd scylla
    cat ../backup_files.out | awk '{print $2}' | sort | uniq | xargs mkdir -p
  3. Download the SSTables to the created directories

    cat ../backup_files.out | xargs -n2 aws s3 cp
    cd ..
  4. Start the Scylla container with mounted data volume. Make sure to use the same Scylla version used to create the backup.

    docker run --name scylla -p 9042:9042 -v `pwd`/scylla:/var/lib/scylla/data --memory 1G --rm -d scylladb/scylla:4.3.0
  5. Extract schema:

    cqlsh -e "DESCRIBE SCHEMA"  > db_schema.cql