Load balancer endpoint new release bug (loses outside connection)
The server logs print:
"GET /ping HTTP/1.1" 200 OK
"GET /ping HTTP/1.1" 200 OK
...
But when you send a curl /ping request to the HTTPS endpoint it does not go through to the server. The metrics just shows failed requests but no further data.
To gain access again, I have to set all the active workers to 0. Wait for all instances to be idle, and then set the active worker count again.
To reproduce:
- Create a simple ping Docker container with Tag V1, and start the endpoint and set active workers to 1
- Create a new Docker container with Tag V2, and create a new release
- When the new release is idle, terminate the active worker from V1
- The V2 worker access starts and the internal logs print ".... GET /ping HTTP/1.1" 200 OK " but there is no outside access (curl /ping request to the HTTPS endpoint fails)
- Set active workers to 0, wait for it to be idle, and then set active workers to 1, now there is outside access (curl /ping request to the HTTPS endpoint succeeds)
"GET /ping HTTP/1.1" 200 OK
"GET /ping HTTP/1.1" 200 OK
...
But when you send a curl /ping request to the HTTPS endpoint it does not go through to the server. The metrics just shows failed requests but no further data.
To gain access again, I have to set all the active workers to 0. Wait for all instances to be idle, and then set the active worker count again.
To reproduce:
- Create a simple ping Docker container with Tag V1, and start the endpoint and set active workers to 1
- Create a new Docker container with Tag V2, and create a new release
- When the new release is idle, terminate the active worker from V1
- The V2 worker access starts and the internal logs print ".... GET /ping HTTP/1.1" 200 OK " but there is no outside access (curl /ping request to the HTTPS endpoint fails)
- Set active workers to 0, wait for it to be idle, and then set active workers to 1, now there is outside access (curl /ping request to the HTTPS endpoint succeeds)
