connection resets, 502 Bad Gateway errors

[lawrence kalinowski]

Since October 11th, 2014 - incidence of connection resets & 502 bad gateway exceptions rose dramatically. We've decreased the frequency of requests to address the connection resets, but have been unable to get the incidence of 502 errors down to zero.

Best practices advice to avoid this error?

Something change API wise on or around 2014-10-11 that we need to permanently adjust our processes around?

thanks