Inquiry Regarding QlikView Server (QVS) Intermittent Crashes and Restarts

[qwebnm]

Dear Qlik Support,

We are experiencing intermittent crashes and automatic restarts of QlikView Server (QVS) in our environment, and would like to request your assistance.

1. Environment Information

• QlikView Version: 12.40.20100.0 (June 2019 SR1)

• Server Memory: 2TB

• OS: Windows Server (version to be confirmed)

• QVS Setup: Single Node

2. Issue Description

• QVS service intermittently terminates and then restarts automatically.

• The following errors are repeatedly logged:

  • “Internal inconsistency, type CPV, detected”

  • “Fiber loop stall detected (doc = …)”

  • “Core dump: Writing core dump file …”

• Core dump (.dmp) files are generated in the path:
  C:\ProgramData\QlikTech\QlikViewServer\CrashDumps

3. Log Analysis (Internal Findings)

• Performance Logs:
  At the time of the incident, the VMLargestFreeBlock(MB) value drops sharply.
  Notably, despite having 2TB of RAM, when VMLargestFreeBlock falls below ~1.5TB, the QVS service restart occurs consistently.
  This appears to indicate memory fragmentation, where large contiguous memory blocks cannot be allocated.

• Event Logs:

  • “Fiber loop stall detected” is triggered when loading certain QVW documents.

  • This is followed by CPV errors and then a QVS restart.

  • Large .shared files and accumulation of server objects have also been observed in connection with some QVWs.

4. Questions to Qlik Support

1. What is the exact cause of CPV errors, and are there any known conditions under which they can be reproduced?

2. Can Qlik Support assist with analyzing the generated core dump files?

3. Could memory fragmentation (VMLargestFreeBlock drop) and the accumulation of shared files be the primary causes of this issue?

4. Specifically, is it expected behavior that with 2TB of RAM, the QVS service restarts when VMLargestFreeBlock falls to ~1.5TB?

5. Are there any recommended temporary mitigation steps (e.g., adjusting QMC settings, Working Set configuration, Object Calculation Time Limit)?

6. What solutions does Qlik recommend for this issue?

We would appreciate your guidance and support in resolving this matter.

Best regards

[marcus_sommer]

Regarding to the following posts I think the most likely reasons are timing/timeout related:

Internal inconsistency, type CPV error in QlikView... - Qlik Community - 1711455

Error: StdServer: Fiber Loop stall detected

AFAIK should the Qlik server have a complete control to the hardware-resources especially the RAM. No sharing with other tools or a VM host distributes resources or NUMA and similar features are enabled. At least to exclude any external impact such measurements should be disabled.

After that a look to the shared files - removing them temporary and/or a repair/cleaning - may a next step.

And of course a bit trial & error by adjusting a few of the timeout and/or sizing settings within the various configurations may also worth an attempt (make BACKUP's before and/or document what was changed !).