- Production cluster: machines has different external IP address. This can be verified by running “curl http://ipecho.net/plain;echo” script.
- https://api.linkedin.com works for some machine while blocking for others.
How to reproduce the issue from a dev machine that has SSH access to the production cluster.
- Python code that invokes the LinkedIn API.
- On dev machine, run ” sudo ssh -L 443:api.linkedin.com:443 production_id@production_ip”, this will route the localhost:443 to api.linkedin.com:443 via a target production machine.
- DNS spoofing such as dnsmasq:
dnsmasq.hosts: “127.0.0.1 api.linkedin.com”
sudo service dnsmaq restart