Hello,
We recently had a Laserfiche Cloud workflow instance remain in a running state for approximately 3 hours without completing or generating an error. Under normal conditions, this workflow completes in approximately 10 seconds.
The workflow itself is pretty simple. Its primary purpose is to generate a custom Process ID for a given Business Process by reading a value from a Data Management table and incrementing a counter on the appropriate row. The table contains fewer than 30 rows and the workflow performs only a read and an update on one row.
After seeing that the instance kept running for a few hours, I immediately "re-tried" the workflow using the same process context and inputs. The new instance completed successfully in its normal 10 second timeframe without any workflow changes.
Because the replacement instance succeeded immediately, I'm trying to understand what types of conditions could cause an individual workflow instance to stop progressing without failing outright.
My questions are:
- What can cause a Laserfiche Cloud workflow instance to remain running for hours without generating an error?
- When canceling a stuck instance and starting a new instance succeeds immediately, does that typically indicate a platform issue, a resource lock, a service dependency issue, or something else?
As a mitigation, I have added an escalation activity that will notify me if future instances exceed the expected runtime. I am also considering allowing this workflow to run concurrently to not block the business process from running since the workflow itself is not mission-critical.
This workflow supports a time-sensitive business process where users expect near-immediate completion. Even short delays are noticeable, and in this case the 3-hour delay caused operational disruption because downstream activities were blocked while the workflow remained in a running state.
I am interested not only in understanding possible root causes, but also in learning best practices for monitoring, resiliency, and prevention of similar incidents in Laserfiche Cloud.
Thanks!