0
Under review

Jobs Stuck Processing

Liam Schulz 10 months ago in UNIFYBroker Service updated by Matthew Davis (Technical Product Manager) 10 months ago 1

Hi,

We have observed jobs such as Connector imports and Link synchronization will occassionally get stuck in a processing loop and not complete. This causes a block in operations as Broker cannot import or synchronize new data. To clear the process a restart has to be performed. Attempting to cancel the job does not have any impact.

This happens intermittently and doesn't appear to have a consistent way of reproducing the issue. I understand this makes it difficult to troubleshoot the issue, so is there other possibilities for a solution we could explore? For example, could there be a timeout introduced so that the job is killed if it runs over a period of time without closing?

Let me know your thoughts and feedback.

Thanks,
Liam

Under review

Hi Liam,

You're right - without any consistent way to reproduce the issue it does make it hard. However, we can try and narrow down where it might be happening to see if there's a pattern.

- Is there specific environments, customers or operations that are getting stuck? 

- Are there log files that can be supplied for times where a job was stuck, which might show a job starting but never finishing? 

- What does the job report when it is stuck (for example, if an import operation is stuck, is it showing any entities returned?) 

- Is there any consistency on the type of operation, for example always a powershell connector getting stuck, or a particular link + operation combo?

A timeout is definitely an option that we could explore, however I'd like to see if we can find the root cause of the problem, as operations shouldn't be getting stuck. So if we can find the cause of the problem, that'd be better for the future of the product than a bandaid fix.