0

I'm trying to create an AWS DMS replication instance with the basic configuration

resource "aws_dms_replication_instance" "test" { allocated_storage = 5 apply_immediately = true auto_minor_version_upgrade = true availability_zone = var.availability_zone engine_version = "3.6.1" multi_az = false publicly_accessible = false replication_instance_class = "dms.t3.small" replication_instance_id = "dumptos3" replication_subnet_group_id = aws_dms_replication_subnet_group.test.id vpc_security_group_ids = [ aws_security_group.dms.id, ] }

The instance is getting stuck in the 'Starting ...' status.

Are there any logs I can check? I can't see any relevant log groups in CloudWatch Logs. I have read the entire section Working with an AWS DMS replication instance and The troubleshooting section in the AWS DMS documentation, and can't find anything relevant.

Improve this question
3
  • Check Cloudwatch Logs, and I wonder if the small disk is a problem, try increasing it. Also consider if it has internet access or access to appropriate VPC endpoints, I'm not sure what it needs. See if you can increase log levels, but that might be for after the instance is started. There's logs that AWS support has access to as well that users can't access, I spent months working with them with a DMS problem a couple of years back. Type your question into Perplexity or ChatGPT, it'll give you a pretty good answer ;) Commented Jul 2 at 19:57
  • @Tim, I have already spent hours blindly changing security groups, adding VPC endpoints, subnets, tweaking IAM roles and attached policies, searching CloudWatch, CloudTrail and the documentation for anything useful. My ChatGPT history is approaching a short book length, and no, it didn't give me a good answer, just a suggestion to check CloudWatch Logs. There are no logs for the instance. Commented Jul 3 at 14:21
  • It's been a while since I used DMS extensively, I forget what logs it has. Try putting it in a public subnet with an internet gateway to see if it's a dependency it's downloading from somewhere - which isn't necessary as I always had DMS in a private subnet but it's a testing step. If that doesn't work I think it's time to engage AWS support or your AWS account manager. Developer level support isn't that expensive. Commented Jul 3 at 19:38

1 Answer 1

Reset to default
0

A DMS replication instance takes more than a few minutes to start. There is an old GitHub issue that suggests increasing timeouts for the aws_dms_replication_instance resource creation to 15 minutes. The current create timeout is documented as 40 minutes. In my case, it took between 15 and 20 minutes to create an instance or change the instance class, and I was bailing out at the 10-minute mark before discovering that old GitHub issue.

Improve this answer

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.