Recently in a project as we were doing the setup of the system, our hosting provider made a typo in one of the MANY ports that our solution uses and it cause a lot of confussion.
When creating the VIP (Virtual IPs) and ports mapping and the ports got flipped. Our hosting provider gave us a document saying this are your servers:
| IP | server service |
| -- | -- |
| 10.2.3.120 | VIP for live |
| 10.2.3.121 | VIP for auth |
| 192.168.1.140 | IP for live |
| 192.168.1.141 | IP for auth |
But we found out it was reversed, `.120` was auth and `.121` was live.
So after a lot arguing and a lot of hours of explaining what happened, how to test, and reporting to people.
I had the need to find a way to bring down the server and effectively test that we are hitting a particular server in a specific port, not through the ports that are used by the server and services because sometimes they may not necessarily tell me if I am hitting auth. My options:
- Check the access log in the web server
- Create different data in auth and live (a user or a product that exists in auth but not in live)
- Enable accesslog trace in the app server
Find a way to test without the app server layer?
Solution:
From Auth:
1) logon to auth at 192.168.1.141
2) `nc -l 5001`
From a client try:
3) http://10.2.3.120:5000
4) in auth, check the output of the `nc` command
No comments:
Post a Comment