We recently upgraded to 1.63.4 version and many executions are getting stored in execution_entitiy table without startedAt time. Based on the bug fixes list in tag Release [email protected] · n8n-io/n8n · GitHub it is fixed, but actually it’s not fixed and we are facing issue on multiple workflows.
What is the error message (if any)?
Please share your workflow
(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)
Hey Chetan, can you please check the createdAt on these executions to see of these were created in the past, or if such executions are still being created ?
can you check in the database whether the errored out executions have a stoppedAt timestamp and the actual execution data in the database?
Also, if you could share one or two of those workflows where this has been happening - that would be helpful to see if there’s any pattern in particular nodes.
hey @Chetan_Jain our current suspicion is that this was being caused by the lack of a transaction in the code that creates the execution_entity + execution_data in the DB, which was preventing these executions from ever starting.
We’ve pushed a fix to start using a transaction around this code now, and backported this to 1.64.3 and 1.65.1.
Hi @netroy Thanks for fixing. As you have deployed these changes on a newer version than currently we upgraded to, so we will check and update you soon on the same.
Hello @netroy
In the new version 1.64.3, all executions are entering a queued state. For these queued executions, the following timestamps are consistently set to null, even after an extended period (over 4 hours):
@Chetan_Jain@Manish_Dhanwal Thanks for the update. Are you able to share a minimal version of the workflow that’s causing the issue? I have not been able to reproduce this so far.