This message was deleted.

This isn't really a Pulumi specific issue, it's just that policy attachement is slow in aws. We have tests that just are calling the aws api directly to set up policies and they just have to spin wait for about 10 seconds until the policy starts to work.

Added a wait time of 60 seconds and more .. still no luck .. and attachment is happening immediately after the error .. if I retry the run , it’s runs smooth ,with out an error .

creating some environment which require the role with the policies

Are you trying to change the role pulumi uses to create aws resources halfway through the program?

Yes , creating it

What language?

Python

Are you making a new aws.Provider with configuration for the role?

Nope , will try that ..I was using rolepolicyattachment

Try making a new provider, that will tell pulumi to use those credentials instead

Provider didn’t solve the problem

You might need to put a sleep in, I'm not sure the aws provider will retry on credential failure