Under review

UNIFYBroker GUI reporting service unavailable intermittently

Hayden Gray 7 months ago updated by Beau Harrison (Senior Software Engineer) 6 months ago 5


Not sure if this has been reported before, but couldn't see anything on my quick look. I am currently in an environment running Broker v5.3.1 Revision #4, and am experiencing what appears to be a UI bug where randomly the web page will begin to display "Service Unavailable" all over the page.

After several seconds to a minute or two, the page will then flick back to normal and continue in this pattern. Some days it will happen more often than others, with no increase in job frequency. Occasionally a service restart will improve the frequency of it occurring, however it will slowly return over time.

I've investigated a bit and it appears to not noticeably affect the current running jobs in UNIFYBroker, nor importing running from Broker into MIM. I have also see it happens more frequently when jobs are running in Broker, but not attributed to any one job in particular. There are also no errors in the Event Log, however there are some in the UNIFYBroker log, though they don't appear to be UI specific, nor can I say I have seen these on every occasion this happens.

The error I'm referring to in particular is:

"Unable to raise complete notification:

System.ApplicationException: This operation returned because the timeout period expired. (Exception from HRESULT: 0x800705B4)
at System.Threading.ReaderWriterLock.AcquireWriterLockInternal(Int32 millisecondsTimeout)
at System.Threading.ReaderWriterLock.AcquireWriterLock(TimeSpan timeout)
at Unify.Framework.Collections.ThreadsafeObjectQueueBase`2.Push(TItem item)
at Unify.Framework.Notification.TaskNotifier`1.Dispose()

If you need any more information just let me know.


Affected Versions:
Fixed by Version:
Under review

Hey Hayden,

Are you running the UI component through the inbuilt web server or through IIS? If you're not using IIS, can I recommend configuring that to see if it helps: https://voice.unifysolutions.net/knowledge-bases/7/articles/2943-configuring-unifybroker-for-use-with-iis 

As of broker 5.2, the inbuilt server was deprecated.

Also, are they running on the same server as each other or different servers?

The UI talks to the service over a REST api, and does regular updates through background jquery updates. So if the browser can't get to the backend service, that's when the errors would show. That wouldn't normally affect service functionality - unless the backend service is actually becoming unavailable.

Next time it happens, either check the web browser logs (using the developer console) or use some network tracing tools and refresh the page to see if it's returning any specific errors or if its just timing the connection out between the UI and the API. You can also test pinging the API endpoints through a script or postman style tool.

Hi Matthew,

Sorry I forgot to mention above. The service historically has been running self hosted while this has been occurring, however I did switch it over to IIS a couple of months ago and it still appears to be occurring.

And yes this is all happening on the one server. So the service and the hosting is setup on the same machine.

The service has just been restarted today so I'll keep an eye out and do as you suggested when it happens again.


Hi Hayden

There's not much to go off in the browser logs as the issue is with communication between the web service (running on IIS) and Broker, not IIS and the browser. It would be helpful if you could capture this. This traffic is on port 59990 so Fiddler would need to be configured to capture on this port, or use Wireshark/rawcap instead.

Regarding the server, is it running any other services other than Broker and IIS? Does it meet the required specs for Broker plus whatever else is running?

Ok Thanks Beau, I'll see what I can organise with the customer so next time it happens I can use rawcap or something similar to capture some traffic.


Yeah, that fine. More than enough to not be responsible for this issue.

Looks like that trace wasn't done on the loopback network adapter. Doesn't contain any of the Broker web requests on localhost. It's not super important until the service unavailable error happens again, anyway.