This system is used to connet to a remote server behind a firewall using reverse ssh tunnel.

• client machine
• server machine
• shared queue

The client (publicly accessible) will send a request for tunnel creation to a shared queue (Amazon SQS queue) since it cant access the server directly.

The server listens forever to the shared queue. when a request received, the server creates a reverse SSH tunnel to the client and will map one of its TCP ports to the client machine.

The client can then connect to that port using localhost:port

The server can access the client (ie. it can authenticate using ssh keys).

$ ssh user@client_ip

set environment variables by edit config file. Those vars used by the server to fetch messages from a shared queue

export AWS_ACCESS_KEY_ID=XXX
export AWS_SECRET_ACCESS_KEY=XXX
export AWS_DEFAULT_REGION=XXX
export QUEUE_URL=XXX

source the config file and run the server side. the servers will start poll the queue

$ . config
$ ./pytunnel.py --d

Edit the request file chen-work.json NOTE:

• ip can be 'dynamic'. this allow the requester host to inject its own public ip

source config and request a tunnel

$ . config
$ ./pytunnel.py --r chen-work.json

1. change server side --d to accept chen-dev39.json and not pnly chen-dev39 (then bash completion helps.)