Schedule Trigger node triggering multiple instances at the same time

Questions
1

Describe the problem/error/question

In my workflows which execute on a fixed schedules/intervals using Schedule Trigger node lately I am facing an issue where at the same time multiple instances/triggers gets fired up. Snapshots below:

image
image302×710 26.7 KB

image
image392×812 21.1 KB

This workflow is supposed to run every 10 minutes, and it does. But the problem is it runs multiple times at the 10th minute. It does not have a pattern - it runs once, twice, or even thrice at the same time.

If anyone has encountered a similar type of issue, your guidance will be really appreciated.

Information on your n8n setup

  • n8n version: 1.50.1
  • Database : PostgreSQL (Cloud SQL)
  • n8n EXECUTIONS_PROCESS setting : own (default)
  • Running n8n : Self Hosted on Google Cloud Run
2

Hi @devuser2 switch your schedule trigger to custom cron, and there set your time:

and then let me know if that does not get triggered, ALSO make sure your flow is published/Active.

3

Thanks for replying @Anshul_Namdev
I was previously using custom cron before I changed it to this and previously also the behavior was the same. So I believe changing it again won’t work. Also, the workflow is already active.

4

@devuser2 i think you should upgrade your n8n instance from v.1.50.1 to 2.x.x and i think that would resolve the issue.

5

hi @devuser2
I’ve seen that the best way to handle this is running the main n8n instance with only one replica for webhooks and the scheduler (max-instances=1). Then, you scale processing separately using queue mode with workers. Since the Schedule Trigger needs to live in a single leader process, this setup prevents duplicate executions while offloading the actual workload.

7

This is an issue with the n8n Cloud run. This is because n8n can spin up multiple container instances each with its own scheduler which causes duplicate trigger fires.

A potential quick fix could be to set the Cloud Run to a single instance (max-instances=1). The best fix for the long term is to switch the executions to a queue, ‘EXECUTIONS_MODE=queue’, so only one scheduler is in charge.

Also might be a good idea to upgrade from 1.50.1 as there have been scheduler improvements since that version.

8

Thank you for your replies. I think I’ll try upgrading my n8n first. If that still doesn’t fix this I’ll try changing execution mode to queue with workers.