This is a complete head-scratcher.
I've got a Rails application running on Passenger.
I've tested the services from several different clients (via curl) and found them performing perfectly. EXCEPT for one particular client (which just happens to be THE client) where we observe POST requests are failing intermittently (about 60-70% of the time).
The failed requests show up in the Apache access log with a 500 response but the application code is not even invoked.
curl reports something like this:
The response is:
Done waiting for 100-continue
< HTTP/1.1 100 Continue
* Recv failure: Connection was aborted
* Closing connection #0
curl: (56) Recv failure: Connection was aborted
Also observed "Gateway timeout" error messages via Charles Proxy.
The failure rate seems to be higher when sending more data.
I'm really baffled by the error because it only shows up for a particular client. I can only guess there's something in the network route that is causing the problem, but haven't the slightest idea how to fix it.
Anyone seen something like this before?