Locking v6.0.1

To prevent conflicts between various operations in the cluster, PGD uses a distributed locking mechanism to ensure the only one node can perform a specific operation at a time.

This is particularly important in a distributed environment where multiple nodes may attempt to modify the same data concurrently. As PGD Essential is a single-write node cluster, it does not have to deal with distributed locking in the same way, as there is only one node that can perform write operations at any time. PGD Expaned, however, has multiple write nodes, and so it must will always use distributed locking to ensure integerity.

Kinds of Locks

PGD uses several kinds of locks to manage concurrent access to data and resources in the cluster:

DDL locking

DDL (Data Definition Language) locks are used to manage access to database objects such as tables, indexes, and schemas. When a DDL operation is performed, a lock is acquired on the object being modified to prevent other operations from interfering with the change. This ensures that the structure of the database remains consistent and prevents conflicts between concurrent DDL operations. Read more about DDL locking in the DDL Locking reference documentation.

DML locking

DML locking is closly related to DDL locking and is used to add an extra layer of protection to a DDL operation being replicated by also halting any DML operations that would conflict with the DDL operation. Again, this is only needed in a multi-write node cluster, and is not used in PGD Essential.

Which locks are used when?

The locks used in PGD depend on the type of operation being performed and the configuration of the cluster. In general, DDL locks are used for schema changes, while DML locks are used for data modifications. A full list of the locks used in PGD can be found in the DDL command handling matrix documentation.