Overview

Currently there is a “single thread” pool for operation chunks preparation/processing and configurable “multi thread“ pool for chunks. This means we can create multiple migration operations for multiple tenants but they will be processed sequentially one by one. While in a scope of operation - chunks will be processed in parallel using “multi thread“ pool.

Operation level thread pool was limited to 1 thread to avoid folio execution context mixing which is caused if thread limit increased to operation level pool.

Contexts mix observed in logs, f.e. we see one tenant in log message header and another tenant in database schema name of log message body. So datasource got wrong tenant somewhy

Requirements/Scope:

1. Come up with a solution to allow multiple migration operations for different tenants to run in parallel without contexts being mixed

Approach - TBD

See

Additional point: save files on aws in the end of each chunk processing