Multi-node setup to run the migrations

scripts/add_nodes.sh

@@ -3,7 +3,7 @@
 if [ -z "$1" ] || [ -z "$2" ] || [ -z "$3" ] || [ -z "$4" ] || [ -z "$5" ]
   then
     echo "Env variables not provided"
-    echo "Usage: ./ipfs_postgres.sh <MIGRATION_FILE_PATH:string> <NODE_NAME:string> <NODE_HOST:string> <NODE_PORT:numeric/string> <NODE_DATABASE:string> <EXECUTE_SQ:bool>"
+    echo "Usage: ./add_nodes.sh <MIGRATION_FILE_PATH:string> <NODE_NAME:string> <NODE_HOST:string> <NODE_PORT:numeric/string> <NODE_DATABASE:string> <EXECUTE_SQL:bool>"
     echo "Only <EXECUTE_SQL> is optional"
     exit 1
 fi

scripts/init-access-node.sh

@@ -0,0 +1,29 @@
+#!/bin/sh
+set -e
+
+# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-config/
+
+# To achieve good query performance you need to enable partition-wise aggregation on the access node. This pushes down aggregation queries to the data nodes.
+# https://www.postgresql.org/docs/12/runtime-config-query.html#enable_partitionwise_aggregate
+sed -ri "s!^#?(enable_partitionwise_aggregate)\s*=.*!\1 = on!" /var/lib/postgresql/data/postgresql.conf
+grep "enable_partitionwise_aggregate" /var/lib/postgresql/data/postgresql.conf
+
+# JIT should be set to off on the access node as JIT currently doesn't work well with distributed queries.
+# https://www.postgresql.org/docs/12/runtime-config-query.html#jit
+sed -ri "s!^#?(jit)\s*=.*!\1 = off!" /var/lib/postgresql/data/postgresql.conf
+grep "jit" /var/lib/postgresql/data/postgresql.conf
+
+# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-auth/
+# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-auth/#password-authentication
+
+# Set password_encryption = 'scram-sha-256' in postgresql.conf on the access node.
+sed -ri "s!^#?(password_encryption)\s*=.*!\1 = 'scram-sha-256'!" /var/lib/postgresql/data/postgresql.conf
+grep "password_encryption" /var/lib/postgresql/data/postgresql.conf
+
+# Append to data/passfile *:*:*:ROLE:ROLE_PASSWORD
+# This file stores the passwords for each role that the access node connects to on the data nodes.
+echo "*:*:*:postgres:password">>/var/lib/postgresql/data/passfile
+chmod 0600 /var/lib/postgresql/data/passfile
+
+# Add "host  all  all  ACCESS_NODE_IP  scram-sha-256" pg_hba.conf on the data nodes.
+# Skipped. Using default "host  all  all  all  scram-sha-256" for now.

scripts/init-data-node.sh

@@ -0,0 +1,22 @@
+#!/bin/sh
+set -e
+
+# https://docs.timescale.com/timescaledb/latest/how-to-guides/configuration/timescaledb-config/#timescaledb-last-tuned-string
+# https://docs.timescale.com/timescaledb/latest/how-to-guides/multi-node-setup/required-configuration/
+
+# It is necessary to change the parameter max_prepared_transactions to a non-zero value ('150' is recommended).
+# https://www.postgresql.org/docs/12/runtime-config-resource.html#max_prepared_transactions
+sed -ri "s!^#?(max_prepared_transactions)\s*=.*!\1 = 150!" /var/lib/postgresql/data/postgresql.conf
+grep "max_prepared_transactions" /var/lib/postgresql/data/postgresql.conf
+
+# Statement timeout should be disabled on the data nodes and managed through the access node configuration if desired.
+# https://www.postgresql.org/docs/12/runtime-config-client.html#statement_timeout
+sed -ri "s!^#?(statement_timeout)\s*=.*!\1 = 0!" /var/lib/postgresql/data/postgresql.conf
+grep "statement_timeout" /var/lib/postgresql/data/postgresql.conf
+
+# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-auth/
+# https://docs.timescale.com/timescaledb/latest/how-to-guides/multinode-timescaledb/multinode-auth/#password-authentication
+
+# Set password_encryption = 'scram-sha-256' in postgresql.conf on the data node.
+sed -ri "s!^#?(password_encryption)\s*=.*!\1 = 'scram-sha-256'!" /var/lib/postgresql/data/postgresql.conf
+grep "password_encryption" /var/lib/postgresql/data/postgresql.conf

scripts/startup_script.sh

@@ -8,7 +8,7 @@ VDB_PG_CONNECT=postgresql://$DATABASE_USER:$DATABASE_PASSWORD@$DATABASE_HOSTNAME
 # Run the DB migrations
 echo "Connecting with: $VDB_PG_CONNECT"
 echo "Running database migrations"
-./goose -dir migrations/vulcanizedb postgres "$VDB_PG_CONNECT" up
+./goose -dir migrations/vulcanizedb postgres "$VDB_PG_CONNECT" up-to 21
 
 # If the db migrations ran without err
 if [[ $? -eq 0 ]]; then