We need to our nodes to connect to S3 from a single IP address, an elastic IP attached to our gateway node.
So - based on this design, we needed a way to only allow access to a set of buckets from this single IP address. For example purposes we are using the IP of 72.309.38.2.
IF you were to follow the initial example laid out by the AWS documentation - you’ll end up with a policy that probably looks similar to this.
{
"Id": "S3PolicyId1",
"Statement": [
{
"Sid": "IPAllow",
"Effect": "Allow",
"Principal": {
"AWS": "*"
},
"Action": "s3:*",
"Resource": "arn:aws:s3:::my-wicked-awesome-bucket/*",
"Condition": {
"IpAddress": {
"aws:SourceIp": "72.309.38.2/32"
}
}
}
]
}
What you’re going to find, after banging your head on the table a few times, is that this policy does not work. There does not appear to be an implied deny rule with S3 buckets (similar to how IAM access policies are setup). I did my testing with the s3-curl command - When you have that tool setup - you can make a query like the following:
./s3curl.pl --head --id=AKIAYOURACCESSKEY --key=YourSecretKey -- https://s3.amazonaws.com/my-wicked-awesome-bucket/even-more-awesome.file
Using the policy above - returned this:
HTTP/1.1 200 OK
x-amz-id-2: BLPqeibX8nZGnSDNi9zRhb+6a8fDiOW6Ij1OXhadWknJKCX9WAb7x1xNETvdXAEv
x-amz-request-id: F6AF36D912E6003B
Date: Fri, 27 Apr 2012 00:25:52 GMT
Last-Modified: Tue, 17 Apr 2012 17:50:11 GMT
ETag: "8553074962ba71d8b2b600b971ba80a8"
Accept-Ranges: bytes
Content-Type: application/pdf
Content-Length: 1451065
Server: AmazonS3
After troubleshooting this with the AWS support team - they let me know of the following:
By default accounts are restricted from accessing S3 unless they have been given access via policy. However, S3 is designed by default to allow any IP address access. So to block IP's you would have to specify denies explicitly in the policy instead of allows.
Once I learned this - the policy was easy to adjust. I flipped around the policy from allowing access from only my IP address to denying access from everywhere that was NOT my IP address.
My new policy looked like this:
{
"Id": "S3PolicyId1",
"Statement": [
{
"Sid": "IPDeny",
"Effect": "Deny",
"Principal": {
"AWS": "*"
},
"Action": "s3:*",
"Resource": "arn:aws:s3:::my-wicked-awesome-bucket/*",
"Condition": {
"NotIpAddress": {
"aws:SourceIp": "72.309.38.2/32"
}
}
}
]
}
And when I ran my s3-curl command again - I now received the following response.
HTTP/1.1 403 Forbidden
x-amz-request-id: 3B90483F655AA692
x-amz-id-2: 0XEO2miEzZEwbU2N5zPwzLtX4J7BJx/LLXpvSLPpQ7e2wuocx49katsD+ZcqwYA7
Content-Type: application/xml
Transfer-Encoding: chunked
Date: Fri, 27 Apr 2012 00:27:14 GMT
Server: AmazonS3
So - based on this design, we needed a way to only allow access to a set of buckets from this single IP address. For example purposes we are using the IP of 72.309.38.2.
IF you were to follow the initial example laid out by the AWS documentation - you’ll end up with a policy that probably looks similar to this.
{
"Id": "S3PolicyId1",
"Statement": [
{
"Sid": "IPAllow",
"Effect": "Allow",
"Principal": {
"AWS": "*"
},
"Action": "s3:*",
"Resource": "arn:aws:s3:::my-wicked-awesome-bucket/*",
"Condition": {
"IpAddress": {
"aws:SourceIp": "72.309.38.2/32"
}
}
}
]
}
What you’re going to find, after banging your head on the table a few times, is that this policy does not work. There does not appear to be an implied deny rule with S3 buckets (similar to how IAM access policies are setup). I did my testing with the s3-curl command - When you have that tool setup - you can make a query like the following:
./s3curl.pl --head --id=AKIAYOURACCESSKEY --key=YourSecretKey -- https://s3.amazonaws.com/my-wicked-awesome-bucket/even-more-awesome.file
Using the policy above - returned this:
HTTP/1.1 200 OK
x-amz-id-2: BLPqeibX8nZGnSDNi9zRhb+6a8fDiOW6Ij1OXhadWknJKCX9WAb7x1xNETvdXAEv
x-amz-request-id: F6AF36D912E6003B
Date: Fri, 27 Apr 2012 00:25:52 GMT
Last-Modified: Tue, 17 Apr 2012 17:50:11 GMT
ETag: "8553074962ba71d8b2b600b971ba80a8"
Accept-Ranges: bytes
Content-Type: application/pdf
Content-Length: 1451065
Server: AmazonS3
After troubleshooting this with the AWS support team - they let me know of the following:
By default accounts are restricted from accessing S3 unless they have been given access via policy. However, S3 is designed by default to allow any IP address access. So to block IP's you would have to specify denies explicitly in the policy instead of allows.
Once I learned this - the policy was easy to adjust. I flipped around the policy from allowing access from only my IP address to denying access from everywhere that was NOT my IP address.
My new policy looked like this:
{
"Id": "S3PolicyId1",
"Statement": [
{
"Sid": "IPDeny",
"Effect": "Deny",
"Principal": {
"AWS": "*"
},
"Action": "s3:*",
"Resource": "arn:aws:s3:::my-wicked-awesome-bucket/*",
"Condition": {
"NotIpAddress": {
"aws:SourceIp": "72.309.38.2/32"
}
}
}
]
}
And when I ran my s3-curl command again - I now received the following response.
HTTP/1.1 403 Forbidden
x-amz-request-id: 3B90483F655AA692
x-amz-id-2: 0XEO2miEzZEwbU2N5zPwzLtX4J7BJx/LLXpvSLPpQ7e2wuocx49katsD+ZcqwYA7
Content-Type: application/xml
Transfer-Encoding: chunked
Date: Fri, 27 Apr 2012 00:27:14 GMT
Server: AmazonS3
