Hi,
I had an upstream defined in my config with keepalive 60. But the server is a legacy one and does not handle keep alive properly.
So I removed the keepalive attribute and the errors I was seeing on the client from the upstream went away.
But now I see a ton these info log lines:
2017/10/03 04:37:51 [info] 1933#0: *6091340 recv() failed (104: Connection reset by peer) while sending to client, client: 164.40.242.212, server: kong, request: "POST /public-api/v1/fs/Shared/functional_tests_2017_10_03_06_37_49_068 HTTP/1.1", upstream: "http://10.116.4.42:7280/public-api/v1/fs/Shared/functional_tests_2017_10_03_06_37_49_068", host: "wdio.qa-egnyte.com"
I dont see any functional errors in our system, and the info log seems harmless, but I was still curious to understand what this means and if it has any side effects at scale.