Description of problem: As reported in bug 431439 for RH5 (comment 2), the RH4 version of net-snmp returns incorrect data as response to the query for HOST-RESOURCES-MIB::hrProcessorLoad. According to comment 2 of the 431439 bug, and a review of the source code, the net-snmp returns average length of run queue rather than Processor Load. There is a patch to fix this being tested for release with RHEL5. The same fix should be tested and released in subsequent RHEL4 updates Version-Release number of selected component (if applicable): net-snmp as distributed with RHEL4 How reproducible: Every time Steps to Reproduce: snmpwalk -v1 -c public localhost HOST-RESOURCES-MIB::hrProcessorLoad HOST-RESOURCES-MIB::hrProcessorLoad.768 = INTEGER: 2 Actual results: average length of run queue Expected results: Processor Load
The version of net-snmp in RHEL5 and RHEL4 are significantly different, the RHEL5 implementation of hrProcessorLoad can't be easily ported back to RHEL4. And AFAIK there is no customer complaining, so the priority seems to be very low to me.
#1 I am a client and I am raising the concern that the numbers reported to all my monitoring system are incorrect. #2 I have been a redhat subscriber from day 1 and I am extremely concerned over the suggestion that because it is difficult and that no one has noticed the bug, a fix is not worth pursing. The definition of the errant mib is processor load not run queue. This qualifies this is a bug, not a feature request. Unfortunately it's difficult to tell that this is the case unless you dig into the code of the implementation and see it. I, like most people, trust the numbers from the SNMP stack to be accurate. I expect this is the biggest reason it has not been reported before. I would respectfully request as a user and service provider of RedHat systems, that this bug be officially recognized and corrected. My understanding is that this type of fix is covered under the subscription cost for RedHat support.
I am sorry if my previous comment sounded rude and impolite, that was not an intention and I apologize for that. Bugzilla is engineering tool and request which do not go through RedHat support (http://redhat.com/support) have lower priority. If you are RHEL subscriber, please log in there, fill new ticket and you request should get proper attention.
Fixed together with bug #469607. Please note that: - the agent needs one minute after startup to collect enough data to report any hrProcessorLoad - after that, the value is updated every 5 seconds
Release note added. If any revisions are required, please set the "requires_release_notes" flag to "?" and edit the "Release Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. New Contents: * UCD-SNMP-MIB::ssCpu* statistics were calculated on the basis of averages since the agent started. Similarly, HOST-RESOURCES-MIB::hrProcessorLoad was calculated as an average of load since the last query. Neither of these presented a current or useful report of system status. These values are now calculated on the basis of activity over the last 60 seconds, and are updated every 5 seconds. Note: it takes 60 seconds from the daemon's initial startup to calculate the OID values.
An advisory 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 therefore 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/RHBA-2009-0984.html