Don't try to automatically alter system_auth replication parameters

This proven to be dangerous in some migration scenarios, and we need to do
this exactly once.  Instead of trying to do it on every startup and hoping it
will fail if already done, just instruct the user to do it manually after
creating a cluster.

Dockerfiles/Cassandra-2/planb-cassandra.sh

@@ -105,19 +105,4 @@ if [ -n "$DC_SUFFIX" ]; then
 fi
 
 echo "Starting Cassandra ..."
-/usr/sbin/cassandra -f &
-
-#
-# Try to create admin user and drop the default superuser (we don't care if it
-# fails, that would just mean we are not the first one to do that).
-#
-sleep 60
-cqlsh -u cassandra -p cassandra \
-      -e "\
-ALTER KEYSPACE system_auth WITH replication = { 'class': 'NetworkTopologyStrategy' $(echo $REGIONS | sed "s/\([^ ]*\)-1/, '\1': $CLUSTER_SIZE/g") };\
-CREATE USER admin WITH PASSWORD '$ADMIN_PASSWORD' SUPERUSER;" \
-    && \
-    cqlsh -u admin -p "$ADMIN_PASSWORD" -e "DROP USER cassandra;"
-
-# Make sure the script don't exit at this point, if cassandra is still there.
-wait
+exec /usr/sbin/cassandra -f

Dockerfiles/Cassandra-3.0.x/planb-cassandra.sh

@@ -99,19 +99,4 @@ if [ -n "$DC_SUFFIX" ]; then
 fi
 
 echo "Starting Cassandra ..."
-/usr/sbin/cassandra -f &
-
-#
-# Try to create admin user and drop the default superuser (we don't care if it
-# fails, that would just mean we are not the first one to do that).
-#
-sleep 60
-cqlsh -u cassandra -p cassandra \
-      -e "\
-ALTER KEYSPACE system_auth WITH replication = { 'class': 'NetworkTopologyStrategy' $(echo $REGIONS | sed "s/\([^ ]*\)-1/, '\1': $CLUSTER_SIZE/g") };\
-CREATE USER admin WITH PASSWORD '$ADMIN_PASSWORD' SUPERUSER;" \
-    && \
-    cqlsh -u admin -p "$ADMIN_PASSWORD" -e "DROP USER cassandra;"
-
-# Make sure the script don't exit at this point, if cassandra is still there.
-wait
+exec /usr/sbin/cassandra -f

Dockerfiles/Cassandra-3/planb-cassandra.sh

@@ -114,19 +114,4 @@ if [ -n "$DC_SUFFIX" ]; then
 fi
 
 echo "Starting Cassandra ..."
-/usr/sbin/cassandra -R -f &
-
-#
-# Try to create admin user and drop the default superuser (we don't care if it
-# fails, that would just mean we are not the first one to do that).
-#
-sleep 60
-cqlsh -u cassandra -p cassandra \
-      -e "\
-ALTER KEYSPACE system_auth WITH replication = { 'class': 'NetworkTopologyStrategy' $(echo $REGIONS | sed "s/\([^ ]*\)-1/, '\1': $CLUSTER_SIZE/g") };\
-CREATE USER admin WITH PASSWORD '$ADMIN_PASSWORD' SUPERUSER;" \
-    && \
-    cqlsh -u admin -p "$ADMIN_PASSWORD" -e "DROP USER cassandra;"
-
-# Make sure the script don't exit at this point, if cassandra is still there.
-wait
+exec /usr/sbin/cassandra -R -f

planb/create_cluster.py

@@ -371,9 +371,7 @@ def generate_taupage_user_data(options: dict) -> str:
         },
         'environment': {
             'CLUSTER_NAME': options['cluster_name'],
-            'CLUSTER_SIZE': options['cluster_size'],
             'NUM_TOKENS': options['num_tokens'],
-            'REGIONS': ' '.join(options['regions']),
             'SUBNET_TYPE': 'dmz' if options['use_dmz'] else 'internal',
             'SEEDS': ','.join(all_seeds),
             'KEYSTORE': str(keystore_base64, 'UTF-8'),
@@ -534,23 +532,45 @@ def launch_normal_nodes(options: dict):
 
 def print_success_message(options: dict):
     info('Cluster initialization completed successfully!')
+
+    # prepare alter keyspace params in the format: 'eu-central': N [, ...]
+    dc_list = ',\n    '.join([
+        "'{}': {}".format(re.sub('-[0-9]+$', '', r), options['cluster_size'])
+        for r in options['regions']
+    ])
+
     sys.stdout.write('''
 The Cassandra cluster {cluster_name} was created with {cluster_size} nodes
 in each of the following AWS regions: {regions_list}
 
-You can now login to any of the cluster nodes with the superuser
-account using the following command from inside the docker container:
+You should now login to any of the cluster nodes to create the admin superuser,
+using the following command:
+
+$ docker exec -ti taupageapp bash
+
+(docker)$ cqlsh -u cassandra -p cassandra \\
+            -e "CREATE USER admin WITH PASSWORD '$ADMIN_PASSWORD' SUPERUSER;" \\
+          && \\
+          cqlsh -u admin -p $ADMIN_PASSWORD \\
+            -e "DROP USER cassandra;"
+
+Then login with the newly created admin account and change the replication
+settings of system_auth keyspace as shown below.
 
 (docker)$ cqlsh -u admin -p $ADMIN_PASSWORD
 
-From there you can create non-superuser roles and otherwise configure
-the cluster.
+(cqlsh)$ ALTER KEYSPACE system_auth WITH replication = {{
+    'class': 'NetworkTopologyStrategy',
+    {dc_list}
+  }};
+
+You can now also create non-superuser application roles and data keyspace(s).
 
 You might also need to update the Security Groups named {cluster_name}
 (in all regions!) to allow access to Cassandra from your application (port 9042)
 and optionally to allow access to Jolokia (port 8778) and/or
 Prometheus Node Exporter (port 9100) from your monitoring tool.
-'''.format(**options, regions_list=' '.join(options['regions'])))
+'''.format(**options, regions_list=' '.join(options['regions']), dc_list=dc_list))
 
 
 def print_failure_message():