We observed unexpected data reported from sar tool. For example, there are large discrepancy between what sar report and what vmstat report on system utilization. On a system with some workload, e.g, a kernel compile, vmstat report 92/3/5 on user/sys/idle utilization, but sar -u would report something out of chart. The data presented from /proc/stat looks fine, seems to be a problem in the sar tool parsing through /proc/stat data. This is just one problem we observed, can redhat verify all other stats reported by sar to be correct? [kwchen@linux-t11 tpc16]$ vmstat 5 procs memory swap io system cpu r b w swpd free buff cache si so bi bo in cs us sy id 4 0 0 0 620288 175280 1071696 0 0 0 1 1 0 1 0 1 6 0 0 0 743648 174464 990624 0 0 0 956 4245 708 93 3 4 5 0 0 0 744688 174672 995664 0 0 0 1071 4247 696 92 3 5 7 0 0 0 746352 174784 1001200 0 0 0 826 4241 700 96 3 1 4 0 0 0 705328 174912 1010368 0 0 0 1904 4257 685 91 3 6 [kwchen@linux-t11 tpc16]$ sar -u 5 0 Linux 2.4.18-e.14smp (linux-t11) 02/07/2003 11:04:55 AM CPU %user %nice %system %idle 11:05:20 AM all 956.65 0.00 32.35 35.00 11:05:25 AM all 940.80 0.00 33.55 49.65 11:05:30 AM all 973.75 0.00 31.40 18.85 11:05:35 AM all 939.10 0.00 32.10 52.80 ---------- Action by: kennethchen Issue Registered ISSUE TRACKER 15460 OPENED BY INTEL AS SEV 3.
Is this still a problem in QU2 errata?
FROM ISSUE TRACKER Event posted 07-31-2003 02:28pm by kennethchen with duration of 0.00 yes, it is still a problem in AS2.1 QU2. One additional information: EL3-beta 1 seems to have the bug fixed.
Is this fixed in the RHEL2.1 U4 beta?
<from client> No. We checked U4 beta, the bug is still there: 12:58:28 PM CPU %user %nice %system %idle 12:58:30 PM 0 0.00 0.00 0.50 1023.50 12:58:30 PM 1 50.00 0.00 29.00 945.00 12:58:30 PM 2 0.50 0.00 1.50 1022.00 12:58:30 PM 3 33.00 0.00 59.50 931.50
Closing as I'm pushing this through standard support channels.