I am attempting to simulate the operation of a gitlab-runner over a poor link (satellite/iridium) it is configured at approximately 2300ms 15% packet loss and 512Kb/s bandwidth.
Of course this is highly sub optimal but I am attempting to test the worst case for my application. The runner can receive jobs
Checking for jobs… received job=141430
but will result in failure when sending the post request with a ‘timeout awaiting response headers’.
Error response from daemon: Get https://registry./v2/: net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers) (docker.go:119:15s)
I was wondering if it was possible to increase this timeout to allow for poor links, or to allow more aggressive retries etc.
This is using the docker 19.03 executor running on Ubuntu 18.04 interacting with Gitlab 13.2.3 and runner version ‘ubuntu-v13.1.3’