Can we have 'enable_lock_retries' by default in DDL migrations?

Why this issue?

We recommend using enable_lock_retries for DDL migrations on high-traffic tables and there are also docs where we have mentioned we prefer to use lock-retries for all migrations, the aim of this issue is to get opinions on the preferred way of using 'lock_retries' and to find solutions to implement this by default.

Decisions to make:

  1. It's fine to always use 'enable_lock_retries' in the migration, as it's not going to hurt even for not so high-traffic tables.
  2. Identify the existing migration helpers without lock_retry mechanism and include them to avoid explicity covering them in with_lock_retries blocks.

References:

Edited by Prabakaran Murugesan