Bug 118647
| Summary: | kswapd in state R and D load constant at 1+ | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | Red Hat Enterprise Linux 3 | Reporter: | Dirk Gfroerer <dirk.gfroerer> | ||||||
| Component: | kernel | Assignee: | Larry Woodman <lwoodman> | ||||||
| Status: | CLOSED ERRATA | QA Contact: | Brian Brock <bbrock> | ||||||
| Severity: | medium | Docs Contact: | |||||||
| Priority: | medium | ||||||||
| Version: | 3.0 | CC: | chrismcc, nhorman, petrides, riel, w.moser | ||||||
| Target Milestone: | --- | ||||||||
| Target Release: | --- | ||||||||
| Hardware: | i686 | ||||||||
| OS: | Linux | ||||||||
| Whiteboard: | |||||||||
| Fixed In Version: | Doc Type: | Bug Fix | |||||||
| Doc Text: | Story Points: | --- | |||||||
| Clone Of: | Environment: | ||||||||
| Last Closed: | 2004-05-12 01:08:40 UTC | Type: | --- | ||||||
| Regression: | --- | Mount Type: | --- | ||||||
| Documentation: | --- | CRM: | |||||||
| Verified Versions: | Category: | --- | |||||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||||
| Embargoed: | |||||||||
| Attachments: |
|
||||||||
Created attachment 98653 [details]
Output of "top -b -n 1"
Created attachment 98654 [details]
Output of lsmod
Take a look at https://bugzilla.redhat.com/bugzilla/show_bug.cgi? id=115438 I think this was caused by our setting the /proc/sys/vm/inactive_clean_percent too low in the RHEL3-U1 update. Please try "echo 30 > /proc/sys/vm/inactive_clean_percent and let us know it this helps. Larry Woodman We experienced the same problem with a x445 with 16GB ram and changed /proc/sys/vm/inactive_clean_percent to 30 (as described in bug 115438), which actually did help. Because I have read in another bugreport that setting this value to 100 is the solution, I would like to know what the correct value is (does it depend on the amount of ram and if so, what's the rule to roughly calculate the best value?) and how it influences the system? Are there other parameters which default values are not set as good as they could be? Is there a documentation where one can find out more about the magic world of kernel-parameters? "Echoing" 30 did help. The load on both machines dropped to zero. Thanks for the quick help! You want 30 and not 100. I probably told you to set it to 100 to quickly verify that it was causing your problem. In the latest kernel update we changed the default value from 5 to 30 because it was too low for many systems so it will no longer need to be adjusted. Larry Thanks a lot for your immediate reply. However, can you give me a hint where to find info / documentation about the various kernel-parameters or about those, which should / can be adjusted? As far as I've seen, the documentation in the kernel- source /usr/src/linux/Documentation is not really always up to date. I'm setting this to "modified" because the U2 kernel has been changed to use a default inactive_clean_percent value of 30. As for where to find further info about VM tuning, I'll let Larry follow up on that question. I've got this whitepaper that may help out here: http://people.redhat.com/nhorman/papers/papers.html An errata has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHSA-2004-188.html |
From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4) Gecko/20030922 Description of problem: After Upgrading to the 2.4.21-9.0.1.EL kernel (WS) two machines have a constant load of 1.00 when they're idle. Didn't have these issues with kernel-2.4.21-4.0.1.EL. top is showing kswapd being either in state R or D. No other processes seem to be running. Version-Release number of selected component (if applicable): kernel-2.4.21-9.0.1.EL How reproducible: Always Steps to Reproduce: 1. Boot machine 2. Look at the output of top Actual Results: load is constantly at +1 even if machine is idle Expected Results: load is nearly at 0 if machine is idle Additional info: