Async parallel execution

[Maneesh_Behera]

I need help regarding a 4-branched workflow, which needs to run on parallel and async.

Issue: Each branch executes seamlessly individually, however when I run the whole workflow as one, the workflow doesn’t return any output and only a couple of branches react as expected.

Please share your workflow

[Epicop]

Hey! Sounds like a timing or merge issue. When branches run in parallel, N8n waits for all of them to complete before moving forward. If one branch errors silently or hangs, the whole workflow stalls. Check each branch for errors using the execution log. Also, if you are merging branches at the end, make sure you are using a Merge node correctly—it needs all inputs to arrive. Try adding Wait nodes or enabling “Always Output Data” on nodes that might fail quietly.

[Maneesh_Behera]

Hi Epicop!
So to manage time related each worker/sub-workflow has “do not wait” enables. So, i believe if one of the branches (sub-workflow) takes longer than expected it should automatically continue flowing.

Also, I do not believe it could be a merge issue as the flow doesn’t even reach the merge node.
Thanks for your input! Cheers

[ramin1]

besides missing multiple merge nodes. keep in mind that when you branch your flow, each branch will be executed from top to bottom.

for example in the below workflow Http #1 executes first and after a couple of seconds (10 seconds approximately) http #2 executes and then after a couple of seconds, the merge node executes

so these two branches do not execute asynchronously

[Epicop]

With “do not wait” enabled on all branches, they fire and forget, so the main workflow continues without collecting their outputs. If you need the results from all branches, you cannot use “do not wait” because the merge node will never receive data.

Remove “do not wait” and let branches run truly parallel. If timeout is your concern, add individual timeout settings per sub-workflow or use Split in Batches with Wait node between batches to control execution flow without losing outputs.

Check execution logs to confirm sub-workflows are actually completing when fired async.