- Newest
- Most votes
- Most comments
Your AWS DMS migration task appears to be stalled during the target load phase. There are several potential causes for this issue:
-
Database Locking: There might be lock contention on the target database. If another application is accessing the target database and locking the table that DMS is trying to write to, it can cause the migration to stall. This creates delays in writing to all tables.
-
Target Table Preparation Mode Settings: If the "Target table preparation mode" parameter is set to "Do nothing" or "Truncate" without having created primary and unique keys on the target tables, DMS will perform a full table scan for each update, which can significantly impact performance and cause the task to appear stuck.
-
Resource Constraints: The target database might be experiencing high latency due to:
- Backups running during the migration
- Multiple availability zones (multi-AZ) being enabled
- Database logging (such as audit or slow query logs) being enabled
-
CDC Load Balancing Issues: If this is a CDC task, DMS uses a load balancing mechanism that might show 0% progress until enough changes accumulate or until you stop and resume the task.
-
Triggers on Target Tables: If there are triggers on your target tables, they can interfere with the DMS replication process.
To troubleshoot this issue:
- Check for lock contentions on the target database
- Review your task settings, particularly the target table preparation mode
- Temporarily disable backups, multi-AZ, or extensive logging on the target if possible
- Try stopping and resuming the task
- Check CloudWatch logs for additional error messages
- Consider increasing the replication instance size if resource constraints are an issue
Sources
DMS Migration Task(only CDC) for Ongoing Replication Type, Migration progress struck at 0% | AWS re:Post
Troubleshooting target latency issues - AWS Database Migration Service
Troubleshooting migration tasks in AWS Database Migration Service - AWS Database Migration Service
Oracle Endpoint Troubleshooting - AWS Database Migration Service
answered 10 months ago

There are no lock contentions on the database:
There are the idle connections from the DMS instance:
select pid, query_start, state, client_addr, query from pg_stat_activity where datname like 'umbriel_platform_clone'; pid | query_start | state | client_addr | query
-------+-------------------------------+-------+---------------+----------------------------------------------------------- 14773 | 2025-11-08 03:05:30.076949+00 | idle | 10.21.245.232 | select pg_backend_pid() 14774 | 2025-11-08 03:05:30.054908+00 | idle | 10.21.245.232 | show max_identifier_length 14778 | 2025-11-08 03:05:30.927684+00 | idle | 10.21.245.232 | ALTER TABLE "public"."affiliate" ADD PRIMARY KEY ( "id" ) 14779 | 2025-11-08 03:05:31.000501+00 | idle | 10.21.245.232 | DEALLOCATE "_PLAN0x1516ac1ed490" (4 rows)
The task preparation mode is DROP_AND_CREATE
There is no "extensive logging" configured on the target.
Numeron stop and restarts of the task have no effect. Stops do not usually succeed unless the instance itself is rebooted, indicating some kind of deadlock within the DMS instance itself.
There data volumne is literally trivial 11 rows - it cannot be a resource issue with dms.t3.medium instance, surely?
There are literally no additional cloudwatch log messages other than the repeated indicating of a server polling failure which presumably neans something to the C programmer responsible for this mess.