There is something in our ERR_NO_RELAY that is causing this error to be logged and for the squid process to fail on start up. I can't show you the exact content of the file but I can tell you
This same file does not cause any issues other 3 instances of squid that we have running internally.
All instances of squid came from the same VM images so they should be the same.
We are unable to reproduce this issue except on the one box and we are unable to debug more on this box right now because it is running in production. I know these files are interpreted so squid can replace certain values available in the session so it may be that a syntax error caused this issue. That does not explain why we cannot reproduce it on other (virtually the same) images. One difference is that the instance of squid that has the issue was under load when the issue occurred. Any suggestions/insight would be appreciated. thanks!