1

I am trying to get codebuild_local working on my Amazon Workspace, a remote Linux desktop. It seems that there are DNS problems when I run a docker container using the default bridge network. But they only happen with our corporate DNS server. If I set the following daemon.json with our corporate DNS server (this same IP is in /etc/resolv.conf):

{
"dns":["10.X.X.X"]
}

I run the following and get a host resolution error:

curl -I google.com

But if I do the following, it works:

{
"dns":["8.8.8.8"]
}

Could this be a routing problem?

re: Using a different docker network - changing to network host seems to solve the problem for one image, but codebuild_local seems to create several images that I'm not sure how to set to use a certain network. Also, I'd like to understand why I'm seeing the DNS behavior.

Update: We think it may be that the workspace has two interfaces, one connected to our VPC and one connected to the internet, and the docker image is bridging to only one, but I'm not sure how to proceed in adding the VPC one to the bridge.

xdhmoore
  • 111
  • 4
  • It sure *sounds* like a routing issue - either the container hasn't bridged to the right interfsace, or you have a broken route between the Docker host and your corporate DNS server(s) – warren Jan 14 '20 at 15:16
  • The route to the host I think is fine because I can hit the DNS server with nslookup from the host. Is there a way to change the interface the container bridges to (or add another interface)? – xdhmoore Jan 27 '20 at 21:04

0 Answers0