[VirtualMachineResize] host 0 does not have enough capacity

Community Support
1

Hi All,

A user is hitting this error when resizing a VM, infact they are reducing the resource - not increasing.

So i’ve changed the RESERVED_CPU and RESERVED_MEM to be regative values.

If I interpret the below settings correctly, CPU can be 100% overcommitted, memory 50GB overcommitted. This is a fairly big environment so over committing is going to be necessary.

RESERVED_CPU -100
RESERVED_MEM -52428800

Currently the Allocated CPU is 7700 / 5900 (131%)

The issue I can still reproduce the resize error with my testing account. I can resize the VM using my oneadmin group account.

Have I used the RESERVED values incorrectly?

Any advice is helpful. Thanks.

*Note i’ve set the same on the Cluster as well just in case without any effect.

2

CPU values are not percentages but CPU units. One CPU is 100 units. To double the amount of CPU in a 58 core machine you should set:

RESERVED_CPU=-5800

oneadmin users are able to resize because by default administrator actions do not check resource limits.

3

Thanks for the response, that fixed it!

I had tried some large negative numbers just in case - but not that large.

I was expecting a percentage - so perhaps the documentation below is incorrect?

http://docs.opennebula.org/4.12/administration/references/schg.html

4

The wording is not correct. When it says percentage refers to the units. 1 CPU = 100 units.