App builder action is returning 504 Gateway

Question

When invoking a long-running action with --result, the CLI returns a 504 Gateway Timeout after ~60 seconds — even though the action continues running successfully in the background.

However, the unexpected behavior is that the CLI appears to automatically retry the invocation after each 504, resulting in multiple activations being created from a single invoke command. In my case, one command produced 3 concurrent activations all processing the same dataset simultaneously.

How can this issue be resolved ?

Meghana_N · Accepted Answer

Thanks for the response ​@manavluharThis was resolved by implementing the below steps :Layer 1: Same-Container GuardA module-level boolean flag is set to true when the action starts. If a second activation lands on the same warm container, it sees the flag and immediately returns 409 without doing any work.Layer 2: Cross-Container GuardAt startup, each activation queries the OpenWhisk Activations API for any other completed activations of the same action in the last 25 minutes. If one is found, it returns 409 immediately.Layer 1 alone misses activations on different containers. Layer 2 alone can't catch two activations starting simultaneously (neither has completed yet). Together they cover all practical scenarios. Both layers fail-open — if the OpenWhisk API check itself fails, the action proceeds rather than blocking all future runs.

manav · Answer

Good one ​@Meghana_N, This is almost always caused by one of two things:External API Latency: Your App Builder action is making an outbound network request to a third-party service [like an ERP, Adobe Commerceor external database]That downstream service hangs, runs slowly, or fails to respond. Because there is no client-side timeout defined in your code, the App Builder action waits indefinitely until OpenWhisk forcibly kills the process at the 60-second mark, resulting in a 504.If you are using a library like node-fetch or axios inside your App Builder action, strictly define an outbound timeout and try.Unresolved DNS Queries: The domain of the external service cannot be resolved, leaving the HTTP request hanging in a perpetual wait state.If the external service genuinely needs more than 60 seconds to process data such as importing a large massive catalog, you cannot perform this synchronously.

Sign up

Login with SSO

Login to the community

Login with SSO

Scanning file for viruses.

This file cannot be downloaded