MitchDrage opened a new issue, #14411: URL: https://github.com/apache/cloudstack/issues/14411
### problem With vm.stats.increment.metrics=true, the network and disk totals for VMware and XenServer VMs are N times the real value, where N is the number of management servers. On these hypervisors each stats poll returns an average over a fixed window (VmwareResource.getVmStats, CitrixResourceBase.getVmStats), so every management server stores roughly the same value. The accumulated total adds up all of those rows. For example, a VM receiving a steady 100 KB/s with 3 management servers stores three rows of ~100 every minute, so the total goes up by 300 each minute. With one management server it goes up by 100. KVM isn't affected, because the agent returns a delta since the last poll by any management server. ### versions 4.20 and later (found while looking at #14397) ### The steps to reproduce the bug 1. Run 2 or more management servers with a VMware or XenServer host. 2. Set vm.stats.increment.metrics=true. 3. Compare a VM's networkkbsread in listVirtualMachines against the vm_stats rows for that VM, grouped by mgmt_server_id. ### What to do about it? I've worked this out from the code and haven't tested it on VMware or XenServer. #14398 should fix it, since it stops each management server polling every host. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
