Error: StdServer: Fiber Loop stall detected

[Andre_Sostizzo]

You may notice fiber loop errors appearing in the Windows, QlikView and Qlik Sense logs.

Example:

StdServer: Fiber loop (10b74b020) stall detected: last fiber loop iteration started 120917 ms ago.

Fiber loop is an entirely normal error to see and is usually just a sign of a busy server. It may not show any symptoms to the end user.
Fiber-loop stalls indicate a problem with a Windows fiber (a kind of thread) that has taken too long to respond. It shows up in the Windows error logs and the user may experience this a client crash as the session has been closed. If the browser is refreshed then a new session is created.

If the peak memory usage is sufficient to drive the server up towards the upper control limit (High Working Set default is 90%), then one can expect to see some level of degraded performance and unusual failures as the in-engine cache will be mostly drained of memory. The required and recurring re-calculations of results will also use a lot of CPU, and this is also likely the path of explanation for the observed stalls.

Environments

• QlikView all supported versions
• Qlik Sense Enterprise on Windows all supported versions

Cause

Fiber-loop stalls indicate a problem with a Windows fiber (a kind of thread) that has taken too long to respond.

Generally caused by a busy server that may or may not be oversubscribed (causing peaks in resource usage) and/or by slow network performance particularly when working with large apps stored on network device or share.

Resolution

It would be more a case of investigating the overall server performance, size of the Apps (if these are very large and stored on a network device) perhaps there are issues with Network speed/performance. 

Any performance issues should be investigated with their local IT team to help reduce the number of fiber loop errors reported.

See What are fiber loop errors in the QlikView server event log? for more details, as well as related content below.
 

Related Content

• How to start investigating a suspected Qlik Sense Performance issue 

[john_oll]

One link is no longer working and should be remove from the article.

[Sonja_Bauernfeind]

@john_oll Thanks for the assist, John!

I modified the link to point at the correct location 🙂 

[korsikov]

Any updates how to fix fiber loop issue?

[Sonja_Bauernfeind]

Hello @korsikov 

Fiber loop stalls logged by Sense by themselves without performance impact should not be a cause of concern.

If a performance impact is seen (which the fiber loop entire are a symptom of, they are not a cause), then we recommend starting off with How to start investigating a suspected Qlik Sense Performance issue .

[korsikov]

hello @Sonja_Bauernfeind 

This are not only degrade performance. The Qlik Sense application stops responding completely. in the QMC, the Engine service is marked as stopped, but the Engine.exe process continues to consume memory, but the processor does not. Windows service Qlik Sense Engine can not be stopped.

[Sonja_Bauernfeind]

Hello @korsikov 

This sounds like the system needs general troubleshooting. I would recommend logging a support ticket if you haven't done so already.

The Fiber loop itself is another symptom of an underlying issue.

All the best,

Sonja 

[korsikov]

@Sonja_Bauernfeind Thank you for attention to my issue. 

thank you for attention for my issue. I will do more investigation and will create new support case with full information about this issue. Now i found noncaptured exception "ExtendedException: Type '2' thrown in file 'C:\Jws\engine-common-ws\src\AppObject\src\QGenericObject.cpp' in function 'CGenericObject::GetSnapshotObject_impl' on line '1463'. Message: 'Resource not found' and additional debug info: 'AmDqa'" in System_Engine log and in same time fiber loop error mesage.
After few minutes in Service_Repository how message "Command=Check service status;Result=500;ResultText=Error: Unable to connect to Engine"

[kashyapdhar]

Hi @Sonja_Bauernfeind / @Andre_Sostizzo 

This is an old thread, but any luck on this issue. This fiber loop tends to slow down even the QlikView services. We have seen instances of fiber loop coming on the reports which were earlier working fine. The report UI just freeze and user has to either close the instance and restart or we have to restart the QlikView services (during production hours which is not an ideal state to be in) which resolves the issue

The QVS typically is not even overloaded when these restarts happen. Average usage on QVS is ~250-350GB  (1.5TB)

[Sonja_Bauernfeind]

Hello @kashyapdhar 

Fibre Loop messages are not what is slowing down your services. They are a symptom of an existing performance impact, such as resource bottlenecks (disk, CPU, RAM) or other factors such as document design. 

I would recommend reviewing the material we have available on troubleshooting and investigating performance issues or connecting with our userbase in the QlikView management forums.

All the best,
Sonja 

[kashyapdhar]

Hi @Sonja_Bauernfeind 

Thank you for your feedback, I did check the server and couldn't find any major instance of i/o being consumed more than usual or CPU / RAM being consumed heavily. 

This is an issue which comes on UI's which are typically bigger in size (> 1GB)  where fiber loop errors are being looked at. Is there anything specific i need to check ?