Resolving The Problem

First, add the following configuration settings to work around a Java bug:

1. Add the following JVM parameter to the JVM hosting Content Engine (CE):
-DFileNet.WSI.HttpFaultResponseCode=599

2. Add the following JVM parameter on the client (FDM):
-Dsun.net.http.retryPost=false

To do so, add it to DeploymentManager.ini
-vm
C:/Program Files (x86)/IBM/FileNet/ContentEngine/tools/_jvm/jre/bin
-vmargs
-Xms256m
-Xmx768m
-Dsun.net.http.retryPost=false

If IBM HTTP Server (IHS) is between FDM and CE, also add to the CE JVM:
-Dsun.net.http.retryPost=false

Second, set the persistent connections to unlimited to work around a WebSphere bug. This is a connection reset during an HTTP chunked input stream read, and is therefore likely to be caused by this WebSphere APAR.

PI11176: RST CONNECTION IF LEFT 1 OR 2 BYTES UNREAD DURING THE HTTP POST AND CHUNKING
http://www-01.ibm.com/support/docview.wss?uid=swg1PI11176

The current work around for this behavior is to set the "persistent connections" setting in the HTTP Channels of WebSphere to "Unlimited persistent requests per connection".

This can be found in the following location in WAS 8.5.5:

1. Sign in to WebSphere Integrated Solutions.
1. Navigate to Servers > Server Types > WebSphere application servers > server_name > Ports.
2. Click "View associated transports" at WC_defaulthost.
3. Click HTTPQueueInboundDefault.
4. Click HTTP inbound channel
5. Under Persistent connections,
5.A. Check <checkbox> Use persistent keep alive connections.
5.B. Click Unlimited persistent requests per connection.

You should change this setting for both WC_defaulthost and WC_defaulthost_secure.