Incomplete Output Stream on the Next POST After One Fails
What does "java.io.IOException: Incomplete output stream" mean in Workday Studio?
One HTTP POST to an external endpoint fails, and the next POST in the same run fails too even though it is perfectly well formed on its own. The transport reuses the underlying connection between calls, so state left behind by the failed request corrupts the one that follows it.
Quick fix
Clear all headers from any message taken off the error path before it reaches the next outbound step, so nothing from the failed request travels forward on the reused connection.
Permanent fix
Build each outbound request from a fresh message rather than reusing or mutating the previous one, and handle errors in a branch that terminates instead of rejoining the main route. If the endpoint still misbehaves across reused connections, switch the transport to an alternative HTTP client implementation so each call negotiates cleanly.
Troubleshooting
- Prove the second request is good in isolation by running the assembly with the first, failing call removed. If it then succeeds, the payload is fine and you are looking at connection state.
- Check the order of the calls. The pattern is always a request that draws an error response, followed by a request that would otherwise have worked.
- Look at what your error path does with the failed message. If you branch on the error and carry that message forward, its headers can travel onto the next request.
- Inspect the headers actually leaving on the second call, for example by temporarily pointing the step at a request inspection endpoint using dummy data and no real credentials.
- Test whether turning off connection reuse changes the outcome. That separates a header carryover problem from a keep alive problem and tells you which fix to apply.