I noticed an issue in multiple vCloud 5.1.2 environments where VMs would report the System Alert “The Virtual Machine’s memory reservation was modified in vCenter Server and doesnot match the reservation set by vCloud Director. Please stop and restart the Virtual Machine from vCloud Director to fix the problem.” I’ve never noticed this warning in 1.5.x, 5.1.0 or 5.1.1 (nothing was modifying the VMs in vCenter).
The VMs with the problem had been either imported from vCenter or through the vCloud OVF upload process and only affected VMs in allocation models of type allocation pool. The environments had a lot of different variables including pointing to vCenter 5.0 U1 and 5.1 U1. Some of the vClouds were upgraded from 1.5.2 and others 5.1.1. The issue affected both org vDCs created before and after the vCD upgrades. The memory allocations settings in the org vDCs were wildly different (except one) and the setting “Make Allocation pool Org VDCs elastic” was disabled. The only commonality that I could find was that org vDC setting “Memory resources guaranteed” was set to the lowest value of 20%. In most environments the System Alert warning went away if I changed the “Memory resources guaranteed” value to 21%. In some environments I had to raise it to 22%. Setting the value back to 20% caused the alert to come back.
It seems like this is could be a bug in vCD 5.1.2 and may have been introduced with all the changes made to allocation pool although more testing could be done.
I tried to use the API to get the alerts but nothing was returned. You can query the vCD database directly to retrieve all VMs with this alert and clear the alert in the vCD UI:
Display all VMs with the system alert
select * from object_condition where condition = ‘vmMemoryReservationModified’
Clear the alert in the vCD UI
update object_condition set ignore = 1 where condition = ‘vmMemoryReservationModified’