RDS t3.small multi-az capacity in us-east-1

Sometimes it isn't useful to apply EC2 documentation, and certainly not forum postings, to RDS. So the RI might not help with your immediate problem but still will help you save costs over time. (You could also contact customer service, explain why you purchased by mistake, and try to get a refund.)

I suspect what is happening is that you are trying to put the Multi-AZ standby into us-east-1e, which for some reason (likely too full to add racks) doesn't have, and doesn't appear to have ever had, t3.micro instances. The question is, why would RDS try to put the standby node there when it won't allow you to select it manually. If I had to guess I'd say that you probably only have two DB Subnets defined, one for the AZ your primary lives in and one for us-east-1e. So RDS, or more specifically the console, tries to put your standby into 1e and EC2 returns the ICE error.

Make sure you have subnets for at least one AZ other than where your primary lives, and other than 1e. If you don't have one for another AZ, create it and try modifying to Multi-AZ again. You could also try using the CLI and explictly specifying which AZ you want the standby in via SecondaryAvailabilityZone in the modify-db-instance command.

Hal

Ps: If I had to guess this is probably poor behavior in the Console rather than at the core API level in RDS. That is probably where they look at your existing subnets and try to determine the appropriate place to create the secondary. If I'm right, you can use the Feedback link in the Console to tell the development team how their trying to make your life easier actually made it harder.

HalTemp wrote:
I suspect what is happening is that you are trying to put the Multi-AZ standby into us-east-1e, which for some reason (likely too full to add racks) doesn't have, and doesn't appear to have ever had, t3.micro instances. The question is, why would RDS try to put the standby node there when it won't allow you to select it manually. If I had to guess I'd say that you probably only have two DB Subnets defined, one for the AZ your primary lives in and one for us-east-1e. So RDS, or more specifically the console, tries to put your standby into 1e and EC2 returns the ICE error.

This indeed seems to be what was happening. I've added another subnet that supports the t3 class, and we'll see after maintenance tonight how it goes but the console error at least disappeared.

Thanks!