Broken (previously working) workflow shows as active and successful

🔧 Bugs & Issues
1

Hi everyone,
I have a quite long automation workflow that is not working properly but is still marked as active. Even if trying the manual run and watching it state “success”. It is behaving erratically and some parts work while others no (that were working before).

I think something may be wrong at a deeper level.

I built a quite long automation workflow that works on creation for a record. Everything was working as I were adding steps until I added the last steps. At that point the workflow was simply throwing an error that was blank (the error was disappearing simply deactivating the last steps). As a solution, I then extracted the last part and put it in a separated workflow. The detached part seems to work but the main part is not behaving properly. Only a very small part is working (the very first actions of every subpart).

It seems as if at some point something broke and even if there is no sign of it (active and successful runs), the workflow is broken. Maybe the root cause is this blank error I was getting previously.

The run history doesn’t show any problem or throttling.

Some additional questions:

What is the maximum (or recommended) number of actions in an automation workflow?

What does a blank error means?

By the way, as a general question, would you recommend to have a long unique workflow or several smaller ones?

Thanks a lot.

1 Like
2

Hi @R.J.,

to understand your issue it would be helpful to analyze your workflow in more detail, for this we would need a screen recording that shows the affected workflow and the errors.

To your general questions. In principle, we would recommend using smaller flows for better clarity, if only for reasons of modality, as in software development. Sometimes, however, this is not possible, e.g. if you need to ensure that steps are processed sequentially and not in parallel.

The automations in Tape basically work via code generation. This means that all graphical blocks are converted into Java script and then processed via our own API.
You can also test this by converting an action into a script action using turn into in the 3-dot menu.
Longer flows mean that more code is generated and has to be processed. If you have very long flows, our expert users also like to use the script action and write everything in code, which can then be optimized more efficiently by chat GPT.

Empty errors can occur especially when working with code, here it is not always possible for the system to understand the cause of the error, but we are constantly expanding the detection to provide the user with more information.

Best regards
Leo