Workflow runs when executing manually but fails on schedule

We have a workflow that runs every 30 mins, it runs when I click on Execute but fails on scheduling without any error message

running from failed execution throws below

Problem with retry

Cannot read properties of undefined (reading ‘nodeExecutionStack’)

Adding merge after upload fashion cloud and b2b causing the error, if we give the node to split node loop without merge nodes after done are not running.

1 Like

Hey @Shalini,

It looks like you have missed a bit if the template, Can you tell us which version of n8n you are running and how it is deployed. The workflow you have shared also doesn’t match the screenshot.

1 Like

latest version only 0.231.3

How do you have n8n running? Is it using docker, NPM or something else?

We are hosting in AKS(Azure Kubernetes)

Are you using queue mode? Can we also get the full workflow that matches the one in the screenshot.

1 Like

Only on adding merge node after upload b2b and fashion cloud throws error, but when the merge node is deleted and directly giving input from B2B and fashion cloud to split in batch running the workflow but we are not getting teams notification on completion

@Shalini are you also running in queue mode?

How to check which mode its running?

It is something you would have set when configuring it in AKS in the files.

Its running fine when I click execute workflow manually but when it runs on schedule it throws Failed without any error message

We have other workflow which is running on schedule which is working fine so I don’t think anything to do with configuring in AKS

On Schedule

Hey @Shalini,

It is possible that it is related to your workflow but I would still like to know if you are in queue mode, When you are on queue mode the workers would trigger the workflow not the main instance like when you press the run now button so knowing this would help rule out any issues there which is why we ask :slight_smile:

Quick edit… The new message you have shared would also have been handy to have in the original post along with the rest of the template we request.

Can you please tell the steps to check what mode the n8n is running on?

Hey @Shalini,

There are no steps to check, When you set up n8n in AKS and you created the configuration files did you set it up to run in queue mode. It would have involved setting EXECUTIONS_MODE to queue and configuring Redis.

I have also had your workflow running on my local instance and it appears to be ok, So it could be that something has gone a bit strange with your local copy.

Do you get the same issue if you duplicate the workflow?

Also on deleting the “merge node”

workflow runs without any error on schedule, but here teams notifications isn’t working after spilt in batch is done

        - name: EXECUTIONS_MODE
          value: queue

yes its queue mode

Looking at the workflow again that merge node does seem oddly placed and I am surprised it works at all as it is waiting for both inputs before moving on so I would expect that to possibly cause some issues but if it is ok from execute in theory on a schedule it would be ok as well.

As you are in queue mode it could be worth making sure all of your workers are on the same n8n version to rule out anything odd there, It might even be worth checking the worker logs to see if they show any errors.

1 Like

Thanks worker node was never updated and still had 0.214 version where the deployment had the latest, updating the worker to latest fixed the error.
Can you also tell is there any other way to place the merge node or merge node not needed at all, since you said its oddly placed?

Hey @Shalini,

Only there, The bit that was interesting was just the use of it as it typically needs both inputs to fire off but one of them is coming from a loop so I am surprised it was looping correctly.

I suspect though the issue may be that the workflow you are using has a newer version of that node or one of the others and that is causing it to fail on the schedule without running.

Here both will have input because all the files input in “split in batch” to upload in both B2B and fashion cloud and the loop continues, once the last batch ran and done in “split in batch” we are sending teams notification

1 Like