Description of problem: I'm unable to get CPU usage statistics via SNMP for some of my high-uptime systems. This is a good description from the sf.net net-snmp bug database: In UCD-SNMP-MIB there are some ssRawCpu* objects defined under systemStats. I use these for RRD graphics and noticed on one machine after several months without reboot, that ssRawCpuIdle reached 4294967295 (=2^32-1) and stayed on this instead of rolling over and starting with 0 again like Counter32 (the type of ssRawCpuIdle) let me expect. Version-Release number of selected component (if applicable): 5.1.2-11.EL4.6 How reproducible: Consistent Steps to Reproduce: 1. Let a system accumulate more than (2^32)-1 timeticks for any field in /proc/stat 2. Query the ssRawCPU* OID in the UCD-SNMP-MIB that corresponds to said field Actual results: From the monitored system: [billm@db03 ~]$ cat /proc/stat cpu 2626530060 574 526603680 14709589002 19217576 9007617 54236522 cpu0 245751702 84 60232005 1926186933 2584965 1240907 7151544 cpu1 392341845 77 67960043 1772887651 2266862 1054551 6637095 cpu2 252492463 86 63314902 1916844755 2553494 1091744 6850682 cpu3 422714591 87 71430603 1739241738 2211938 1069542 6479625 cpu4 245179838 58 60358959 1926633232 2554170 1264439 7157427 cpu5 392748199 63 68429791 1771917241 2287390 1079375 6686065 cpu6 252598705 59 63468155 1916580745 2548675 1122739 6829045 cpu7 422702713 55 71409219 1739296705 2210079 1084317 6445036 intr 18683487120 ctxt 251235855639 btime 1134081588 processes 197892051 procs_running 3 procs_blocked 0 From the monitoring server: [billm@netmon1 ~]$ snmpget -v 2c -c xxxx db03.lulu.com .1.3.6.1.4.1.2021.11.53.0 UCD-SNMP-MIB::ssCpuRawIdle.0 = Counter32: 4294967295 (Note 4294967295 is the max value for a 32-bit unsigned int) Expected results: UCD-SNMP-MIB::ssCpuRawIdle.0 = Counter32: [14709589002 & 0xffffffff] Additional info: http://sourceforge.net/tracker/index.php?func=detail&aid=1197183&group_id=12694&atid=112694 This has been fixed in version 5.1.3 of net-snmp, the patch is small and should be easy to backport to the RHEL release of 5.1.2. The problem exists for the context switch and interrupt OIDs, and there is a patch for those, too, attached to the sf.net bug report.
Created attachment 134923 [details] Fixes the 32bit rollover problem for ssCpuRaw* Lifted from http://sourceforge.net/tracker/index.php?func=detail&aid=1197183&group_id=12694&atid=112694
Created attachment 134924 [details] ctx-intr-overflow for 5.2.1.2, after vmstat.32bit-overflow applied Lifted from http://sourceforge.net/tracker/index.php?func=detail&aid=1197183&group_id=12694&atid=112694
*** This bug has been marked as a duplicate of 223775 ***