What happens when a source endpoint database is restored that is part of a data migration task

Question

Using AWS DMS there are source endpoints (On-Prem SQL Servers) and associated target endpoints (S3).

Could someone with experience tell me what happens when a source endpoint (sql-server) gets restored, and CDC is re-enabled on that database.

For the associated data migration task, I hope that DMS will see this a requiring a new snapshot/initial load (in transactional-replication parlance). If not, how does one contend with this. These restores can happen with databases from multiple servers and on a frequent basis.

Normally, I would test this out, however, our dev accounts are locked down with limited console access, everything has to be scripted in terraform and this would take some time to get a "quick test" up and running.

subhashr · Answer

for CDC DMS relies on DDL/DML transactions captured in source database transaction log. This changes are populated when we enable ms-cdc. As the source db is getting restored the chain of transactions tracked by lsn (from active transaction log) is lost. Thus you need to restart the DMS task so that initial load is reinitiated.

What happens when a source endpoint database is restored that is part of a data migration task

Relevant content