just noticed that the shift swap data action is timing out in architect when calling the wfm endpoint. using sdk 3.2.0 and getting a 504 gateway timeout even though the api call itself returns 200. is this a known issue with async handling or do i need to adjust the timeout settings in the flow?
{
“timeout_ms”: 120000,
“retries”: 3,
“retry_interval_ms”: 2000
}
the default architect data action timeout is too short for wfm endpoints under load. you are hitting the 504 because the platform drops the connection before the async wfm process finishes, even if the api returns 200 internally. increase the timeout in the data action config to 120 seconds. also, wfm apis have stricter rate limits than standard platform apis. if you are testing concurrent shift swaps, add a retry logic with exponential backoff. jmeter tests show that wfm endpoints throttle after 50 concurrent requests per org. check your thread group settings. if you are hitting the websocket limit, the data action will fail silently before the timeout. monitor the api throughput in real time. this is not a bug, just capacity planning.