Can't use General HTTP Provider: AWS_CONTAINER_CREDENTIALS_FULL_URI is not HTTPS and is not within loop back CIDR: [fd00:ec2::23]

[artklochkov]

https://github.com/aws/aws-sdk-cpp/blame/f5309cf6692124b6a20564615a8dab30e974e9d3/src/aws-cpp-sdk-core/source/auth/GeneralHTTPCredentialsProvider.cpp#L36

Looks like the problem is that in case of IPv6 authority part of URI is [fd00:ec2::23], with IPv6 inside [], while in this comparison we are trying to compare it with loopback address fd00:ec2::23, without []. So it leads to an error

Can't use General HTTP Provider: AWS_CONTAINER_CREDENTIALS_FULL_URI is not HTTPS and is not within loop back CIDR: [fd00:ec2::23]

When using with EKS Pod Identity agent, since it's URL (which couldn't be changed with configuration) is:

http://[fd00:ec2::23]/v1/credentials

[sbiscigl]

@artklochkov / @rajanyadav0307 not sure if yall are on the same team working on this but took at look at @rajanyadav0307 's PR and left a comment. This is actually identical to a similar issue we saw a while ago but just for ipv6 endpoints in general. This introduced a new API called GetHost which is the correct RFC 3986 encoding. so in the PR if you just swap GetAuthority use for GetHost I will approve.

I can also push a updated PR if you are unable to make update to yours

[rajanyadav0307]

@sbiscigl ,
Thanks for pointing it out. I think it makes more sense to useGetHost instead of calling GetAuthority as you stated.

I see the code handles for IPv6:

I will update the PR.

[rajanyadav0307]

@sbiscigl ,
I hope we don't need a new PR, last PR should be good to merge?