AWS EC2 – Fix API Error Labels in AWS Management Console

I had the same issue, even contacted support and they could not provide an answer. Exporting credentials to environmental variables does not work for aws s3 ... commands. You have to either use the aws configure util or create a ~/.aws/config file like this one:

[default]
aws_access_key_id=foo
aws_secret_access_key=bar
region=us-west-2

Please reference AWS documentation at http://docs.aws.amazon.com/cli/latest/topic/config-vars.html

Not only did I find it failing to execute at all with the environment exports on EC2 launch via the User Data but also post launch from the command line on a fully deployed running instance.

Unfortunately I cannot comment with my rep so I'll post this as an answer.

I personally would take the same approach as @prateek61.

It's very difficult to diagnose this issue without being able to log in to your server. Linux is very good with not letting people in once the some configuration has changed or been triggered.

Since in AWS there is not console access, this is how I would investigate this assuming that you cannot shutdown the server:

If you can shutdown the server then you can jump to step 3 however I assumed that this is prod and you can't take it down.

1. Create a snapshot of the volume
2. Create a volume from the snapshot you have just took.
3. Mount the new volume on a good working instance where you are able to ssh to.
4. Once mounted go to the volume and investigate:
   • file permissions for the ssh keys
   • the keys itself are they correct
   • check sshd configuration, sudoers etc.
   • check the logs files etc.
   • check if you have firewall running on that server (as this is mounted volume you can't check this however you can check the /etc/rc3.d to see if your the symlink to your firewall is there)
   • sometimes fail2ban or other similar software is running and preventing you from accessing the server again check what is starting with the system in /etc/rc3.d.

After you find what is wrong you will need to figure out how to apply it to your running server i.e. swap volumes, create a new instance, re point the traffic etc. this depends on the purpose of the server and how much downtime you can afford and is a matter for another subject.

Once again I'm posting this as answer as I was not able to comment. If you want to downgrade this, please be king enough to provide a reason.