[IDEA] Transformer Logging Not Consistent

[joshmc82]

Describe the bug
Source and destination transformers do not log to the Server Log console (or mirth.log on the server). This presents challenges when integrating with log aggregation products.

To Reproduce

1. Create a new channel with any data types (I chose RAW)
2. Add a source transformer step with javascript
3. Add code to manually throw an error throw "source transformer error"
4. Send any message to your channel
5. Observe there is no log entry on your server or your Administrator Server log tab.

This does not happen on response transformers. If you put the throw in your response transformer, it will display in the server log.

Expected behavior
ALL transformer steps should log an error in the server log. The behavior across transformers should be consistent.

Actual behavior
No entry gets added to the server log.

Screenshots
Sample channel attached

Environment (please complete the following information):

• OS: Linux (CentOS/Rocky Linux, Alpine)
• Java Distribution/Version: OpenJDK 11, Java 8
• Connect Version: 3.10.1, 3.12, 4.0.1, 4.1.1

Workaround(s)
Have to wrap every JS Transformer Step in a try/catch block and add a logger.error statement to log out the exception.

Example Channel Attached (modified version of channel from jonb to demonstrate the issue)
Pruner Experiment.zip

[tonygermano]

Not sure if I agree that all transformers should go to the server log. These are channel/message errors, and they get logged with the message. It could make for a very busy server log with no good way to filter what you don't want to see. Transformer errors are also more likely to unintentionally contain PHI, which many people would not want going to the server logs.

@pladesma can you share more about the Fix that was committed next week and how this issue was addressed?

[narupley]

This is a good point. We may want to hold off on this for 4.3 and think of a better solution.

We probably want to make this consistent, but the other way actually, where none of the message-specific scripts send errors to the server log when an exception gets thrown. The deploy/undeploy scripts still should, because those are not message-specific.

And then, perhaps we could add an option in the channel settings so you can decide whether or not you want message script-thrown errors to be sent to the server log.

Of course this would be less of a problem if we had better ways to filter the server log in the UI, but that would be a larger and separate issue.

[joshmc82]

I like the idea of it being a setting we could toggle on and off by channel. It could even be off by default to prevent any unintentional logging.