Created attachment 1330554 [details] Screenshot of alert definition Description of problem: I have a control alert defined to trigger on CPU utilization > 10% for a RHV VM or instance (see attached screenshot). The alert profile is assigned to tagged VMs, one of which is consistently running at around 48% CPU utilization. No alert is triggered. Version-Release number of selected component (if applicable): 5.8.1.5 How reproducible: Every time Steps to Reproduce: 1. Set up a real-time control alert as shown in the attached screenshot 2. Generate a CPU load on a VM managed by the CFME appliance Actual results: no alert is generated Expected results: an alert if generated Additional info: An evm.log is available that shows the output from an automate method called 'test26'. This calls perf_collect_metrics on each of the tagged VMs, and shows that the average utilisation the VM 'vvaldez-atomic-host' is above the 10% threshold that the alert is defined for. The log also shows the debug output: MIQ(MiqAlert#evaluate) Evaluating Alert [CPU Utilisation above 10%] for target: [vvaldez-atomic-host]... Result: [false]
Created attachment 1330557 [details] Extract from evm.log
Aren't we saying that "cpu_used_delta_summation" is a VMware-only value? If so then the human-readable value should indicate the provider-specific nature of the counter, for example "Total CPU usage (VMware)" or "Total CPU time usage (VMware)". Similarly any RHV-specific counters such as "cpu_usage_rate_average" should also indicate their provider-specific association in the human-readable value.
https://github.com/ManageIQ/manageiq/pull/16878
Version: 5.10.0.23.20181106165157_92dd189 Verification Steps: 1) Navigate to Control -> Explorer -> Alerts 2) Select "Configuration", "Add a New Alert" 3) Select "What to Evaluate", "Real Time Performance" 4) Select "Performance field" 5) Verified that "CPU-Time used (ms)" has been changed to "CPU-Total time usage (VMware) (ms)"