The pipeline has a hardcoded list of domains that must resolve to different IPs in order for jobs to start. Had the Facebook issues yesterday not been resolved as quickly, this would've led to problems as all pipelines would've had to be stopped and upgraded.
Let's move the domain list to the control node instead so it can be adjusted quickly for all pipelines at once if needed.
The pipeline has a hardcoded list of domains that must resolve to different IPs in order for jobs to start. Had the Facebook issues yesterday not been resolved as quickly, this would've led to problems as all pipelines would've had to be stopped and upgraded.
Let's move the domain list to the control node instead so it can be adjusted quickly for all pipelines at once if needed.