I have several components alerting due to high cpu usage. The odd thing is when I look they show well over 100% utilization. How is that possible and how can I fix that so the app reports/alerts correctly?
I apologize for my confusion but it sounds like the alert is working properly (i.e. alert when their is high cpu usage which in this case is well over 100%). The question now is whether the cpu usage is being reported to and interpreted correctly by APM. Is it only occuring on this one box? If not, what's common between the boxes exhibiting this behavior? If so, have you always had this problem and have you tried restarting it?
I am having the same problem with CPU util in the SQL Server 2005 APM template.
i have a support case open for a few weeks now and have been told that it is due to my servers having multi-core processors. I have not been given ETA of a developmental release to fix this issue
forgot to mention. i've kept the APM in place, but modified the APM advanced alert to exclude the "SQL Server 2005" template from alerting. I can walk you through further if needed.
I am seeing similar behaviour in my installation of Orion 9.5 / APM 2.5. This is happening across multiple servers and components, and has been the case for quite a while (existed in APM 2.0 as well).
Our servers are all Windows 2008, and dual CPU, with varying numbers of cores (some are 1 core, some are 2 core and some are 6 core).
Orion NPM seems to be correctly polling the CPU value and averaging the per-core load into its the plain standard NPM CPU graph. Comparing this with CPU usage in performance monitor shows that NPM is getting the right data.
In APM, however, when adding a process monitor (WMI or SNMP, both show this behaviour), the CPU usage shown seems to be double the actual cpu usage of the application (according to task manager).
eg. on a dual CPU, dual core machine (4 cores total), process.exe is showing a steady 25% cpu in task manager. The CPU graph in Orion NPM correctly shows this server at around 25% CPU total. APM shows the component for process.exe steady at 50%.
Similarly, on a SQL 2005 server with 2 x 6-core cpus (12 cores total), our sqlserver.exe process is mostly between 20% and 30% cpu usage (according to task manager), and has a very steady load. Orion NPM reports the server cpu correctly as 20% - 30%. APM reports sqlserver.exe as higher than 100% most of the time.
These differences in reported load between NPM and APM are fairly constant, and are happening against components which have very stable cpu loads (they do not spike much). I would say it is safe to rule out differing timings between APM and NPM as a cause of this.
Yea I am pretty sure for us its adding cores together as well as swap space and physical ram and comparing it to the physical ram amount.
