Creating a Snowflake Sequence in a New Cluster

Before you use Snowflake sequences, each node in your cluster must have a unique snowflake.node identifier. The value ranges from 1 to 1023, and it must be distinct on every node in the cluster.

Setting snowflake.node in a managed deployment

A pgEdge deployment assigns snowflake.node for you, deriving the value from each node's place in the cluster:

• the Control Plane sets snowflake.node from the node ordinal.
• pgEdge Helm charts derive snowflake.node from each node name.
• pgEdge Ansible playbooks set snowflake.node from the node zone.

When you deploy with one of these methods, no further action is needed to configure the node identifier.

Setting snowflake.node manually

For a self-managed or from-source installation, set the value yourself. Use ALTER SYSTEM to assign a unique node number, then reload the server so the change takes effect. The following example sets the node identifier to 10:

ALTER SYSTEM SET snowflake.node = 10;
SELECT pg_reload_conf();

You can also add the parameter to the postgresql.conf file directly. The following line identifies the host as node 11:

snowflake.node = 11

After you edit postgresql.conf, reload the server for the change to take effect. You can reload from psql:

SELECT pg_reload_conf();

You can also reload the server with pg_ctl:

pg_ctl reload -D "/path/to/your/data/directory"

Once Snowflake is configured, you can begin to use Snowflake sequences by first converting them from Postgres sequences.