after reading a thread about ballooning I did a spot check on our hosts and found one thing that seems strange to me.
The host reclaims memory (using a balloon) from one vm but seems to be unsuccessful. The values (MCTL*) seen in the esxtop screenshot are nearly the same since 24h.
There is just a small up and down.
As you can see we have a bit overcommited, but I don't think there is a real shortage of memory because the vms don't use much of it. This is also affirmed by the summary screenshot I think.
How would you interpret this behavior?
Or did I even get anything wrong with ballooning and everything is as expected?
Thanks for any reply,
Does the guest with the MEMCTLSZ entry have a memory reservation set which is smaller than the amount of memory given to the guest, look at the resources tab on the VM. If so, you will see this behavior. We've see this happen after adding memory to a VM, the mem reservation does not change at times. Best bet is to set to unlimited if you are not overcomitting memory.