Trace javascript action exit code instead of user logs #290
Conversation
| @@ -122,9 +122,9 @@ public async Task RunAsync(ActionRunStage stage) | |||
| else | |||
| { | |||
| var exitCode = await step; | |||
| Trace.Info($"Node Action run completed with exit code {exitCode}"); | |||
There was a problem hiding this comment.
Debug or Info? Info means we need to check runner diag log, Debug goes to user log when ACTIONS_STEP_DEBUG=true
There was a problem hiding this comment.
Good point. I do think Debug makes more sense there. I went ahead and also updated it for the container actions since the same reasoning should apply there as wel
|
debug was meant for users to debug their workflows (not action authors to debug return code). trace wasn't meant as a deeper debug. but as long as it's not console out, I'm happy. I can see both sides of the argument. |
|
I think return code is also useful for action consumers who may have provided bad inputs or otherwise misused the action Debug is way more accessible than trace for this imo, if we consider trace as the runners log (vs the users workflow log) which should probably only be checked if theres suspicion of a bug in the runner itself as opposed to a workflow level error Ill merge later if everyones happy with this |
* Trace javascript action exit code instead of user logs * Debug instead of trace
Fixes #283
Consuming actions from the graph should not imply knowledge of their runtime be it javascript, container etc
We use the exit code to determine step failure/success so tracing that info out is still useful for diagnostics