#25 support for > 2 filelocker instances in HA config

[Nick Silkey]

We experience issues when we spin up > 2 instances of Filelocker in an HA configuration behind a reverse proxy. With one of the multiple nodes being a 'clustermaster' (0) and each member having an iterative clustermemberid (1,2,3), their filelocker configurations are the same. With four nodes, we receive bizarre and hard to reproduce 500s on clustermemberids 2 and 3. Killing filelocker on clustermemberids 2 and 3 where only 0 and 1 are viable (and the only instances available via the reverse proxy), the 500s disappear and other bugs experienced in userland are no more. Is there a filelocker limitation as to why > 2 filelocker instances in an HA configuration would behave like this while ~2 behave differently? The 500s run the gamut of css, js, json, etc.