Changing the Cassandra gc_grace_seconds value (on-prem)

The Cassandra database cluster performance may be impacted by tombstone occurrences which results in slower query response times. When this performance degrades the Agile Service Manager installation, the following procedure to set gc_grace_seconds can be used to mitigate this degradation.

1. Log into a server where Agile Service Manager is running.
2. Find the name of a Cassandra container.

   docker ps | grep cassandra

   For example:

   $ docker ps | grep cassandra
   000000000000   nasm-cassandra:3.11.3.62   "/opt/ibm/start-ca..."   5 hours ago   Up 5 hours   asm_cassandra_1
   

   The pod is identified as asm-cassandra-1
3. Run the following command to exec into the container and start cqlsh:

   docker exec -ti {container name} cqlsh -u cassandra -p cassandra

   For example:

   $ docker exec -ti asm_cassandra_1 cqlsh -u cassandra -p
   Connected to topology_cassandra at asm_cassandra_1:9042.
   [cqlsh 5.0.1 | Cassandra 3.11.3 | CQL spec 3.4.4 | Native protocol v4]
   Use HELP for help.
   cassandra@cqlsh>

Repeat steps 4 to 6 for all tables within the janusgraph key space.

4. Verify the current setting of gc_grace_seconds.

   SELECT table_name,gc_grace_seconds FROM system_schema.tables WHERE keyspace_name='janusgraph';

   For example:

   cassandra@cqlsh> SELECT table_name,gc_grace_seconds FROM system_schema.tables WHERE keyspace_name='janusgraph';
   
    table_name              | gc_grace_seconds
   -------------------------+------------------
                  edgestore |           864000
            edgestore_lock_ |           864000
                 graphindex |           864000
           graphindex_lock_ |           864000
             janusgraph_ids |           864000
          system_properties |           864000
    system_properties_lock_ |           864000
                  systemlog |           864000
                      txlog |           864000
   

5. Change the value to 0 (zero) using the ALTER TABLE command:

   ALTER TABLE janusgraph.{table name} WITH gc_grace_seconds = {gc_grace_seconds value}; 

   For example:

   cassandra@cqlsh> ALTER TABLE janusgraph.edgestore WITH gc_grace_seconds = 0;
   

6. Verify the settings have worked.

   SELECT table_name,gc_grace_seconds FROM system_schema.tables WHERE keyspace_name='janusgraph';

   For example:

   cassandra@cqlsh> SELECT table_name,gc_grace_seconds FROM system_schema.tables WHERE keyspace_name='janusgraph';
   
    table_name              | gc_grace_seconds
   -------------------------+------------------
                  edgestore |                0
            edgestore_lock_ |           864000
                 graphindex |           864000
           graphindex_lock_ |           864000
             janusgraph_ids |           864000
          system_properties |           864000
    system_properties_lock_ |           864000
                  systemlog |           864000
                      txlog |           864000
   
   (9 rows)

   Remember: Repeat steps 4 to 6 for all tables within the janusgraph key space.
7. Exit cqlsh:
   Example:

   cassandra@cqlsh> exit
   $