Block until done not running in linked order (Non-AMP engine)

[ABunce]

Hi All,

I have encountered a strange issue with the Block until done tool, the setup I have in the workflow is:

 

When running in the AMP engine, this runs path one first then path three as expected.

However, when I run this without using the AMP engine the process follows and completes path three first. Path one creates the file headers and inserts a blank row into the data, so always needs to be run before the rest of the process appends data. 

I want this process to be uploaded onto our server for users to run, but our server does not have the AMP engine installed (not going to be available anytime soon). Any suggestions or tool settings I can look at to get around this issue?

Only thought I have is to completely re-create this process tool by tool in the correct order and see if that works.

Thanks,

Adam

[WirkKarl]

Interesting catch, Adam. It definitely sounds like a sequencing issue due to the engine differences. Without AMP, the legacy engine can sometimes handle parallel paths less predictably, even with Block Until Done. You might try restructuring with controlled dependencies using additional dummy tools (e.g. formula or message tools) to force linear execution before the append logic.

On a related note—if you're building workflows for data routing or reporting automation, especially for marketing or lead distribution use cases, check out Phonexa. Their platform integrates process automation with lead management, and can complement Alteryx workflows well for end-to-end data handling.

Hope this helps!

[alexnajm]

@ABunce can you use Control Containers instead to control the sequencing? This would be in place of your Block Until Done (since both can't be in the same workflow) and requires AMP to be on

[caltang]

@alexnajm the OP mentioned they are using it for server without AMP, so control containers are out since it's only compatible with AMP.

Perhaps what I can suggest is for you to review your design process and implement the Block Until Done with the Throttle tool to slow down output #3 to give more time to #1 to finish @ABunce 

[alexnajm]

Ah I missed that @caltang !

[ABunce]

I'm not sure that would work, as the full path needs to complete from the block tool before the next path can begin.

Converting both paths into a Macro and running those through a new block tool would work. It seems to be an issue with the block tool after the process has been updated several times.

[KGT]

Oh... 

This comment gives extra context: "I'm not sure that would work, as the full path needs to complete from the block tool before the next path can begin."

The Block Until Done (BUD) will NOT wait until a stream has finished before executing the next stream. The BUD only controls what happens within it, and so it will hold the data until it has it all, then start releasing down output 1. As soon as the last record has left output 1 (regardless of what is happening on that stream), data will start to be released down output 2 etc. That's how it should work. So, it is entirely plausible that stream 3 executes before stream 1.

The way around this usually involves a batch macro for control.

[apathetichell · Accepted solution]

@KGT ---  throwing a dummy variable in to this could also work. The variable would flow from a step in BUD 1 to an append fields in BUD 3.

Having said this --- I've always found BUD outside of AMP to be MORE predictable than BUD in amp. My memory is that BUD was one of the tools which did not have full AMP compatibility circa 2022. 

A batch macro is 100% the best way to insure timing between your desktop and Server....