So, we are getting the "no live upstreams" error under load. The upstream webnodes have plenty of resources, not exceeding maxclients (Apache), etc. all underutilized.
There are no other interesting errors in the nginx error log that help shed light on this.
Wondering what are all the possible conditions that can cause the "no live upstreams" error, and what recovery occurs, e.g. does the worker give up and another worker retries, etc.
Any help appreciated, as connections to the Apache nodes (upstream) are completely available at the time these occur.
There are no other interesting errors in the nginx error log that help shed light on this.
Wondering what are all the possible conditions that can cause the "no live upstreams" error, and what recovery occurs, e.g. does the worker give up and another worker retries, etc.
Any help appreciated, as connections to the Apache nodes (upstream) are completely available at the time these occur.