Does soft lockup detection cause a 1-second latency in RHEL5?
Environment
- Red Hat Enterprise Linux 5
Issue
- The following statement appears in an HP whitepaper detailing BIOS setting for low latency systems.
- Could you please confirm that Red Hat agree with the statement and have no issue with the recommendation?
-
If you could also add in any specific implications of carrying out the recommendation that would be great.
“Red Hat Enterprise Linux 5 kernel soft lockup detection generates an interrupt every ten seconds. This interrupt results in an approximately one second latency each time it occurs. The default interval is ten seconds. This interval can be extended to a maximum of 60 seconds by writing the appropriate value to /proc/sys/kernel/softlockup_thresh. To minimize the impact of this interrupt on applications, adjust the software lockup detection interval.”
Resolution
- The statement in the HP whitepaper is incorrect. The latency caused by softlockup detection is negligible.
- The latency when a softlockup is actually detected and reported will depend on the speed of the console device.
Root Cause
- Incorrect information in the HP "Configuring the HP ProLiant Server BIOS for Low-Latency Applications" White Paper.
- Relevant Thread: http://post-office.corp.redhat.com/archives/tech-list/2009-November/msg00464.html
This solution is part of Red Hat’s fast-track publication program, providing a huge library of solutions that Red Hat engineers have created while supporting our customers. To give you the knowledge you need the instant it becomes available, these articles may be presented in a raw and unedited form.
Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.
