VMware Cloud Community
aj800
Enthusiast
Enthusiast

Running a virtual appliance and doubled memory to 32GB from 16GB, but now it shows 12GB?

We have a virtual appliance running in our vSphere 6.7 environment (management system for a set of firewalls) and were told by support for it that the recently upgraded version of its OS requires 32GBs of RAM instead of 16 that it had.  At 16GB, performance monitoring showed consistent usage of about 2-4GB at any given time, so he said we should be fine.  But then we had issues deploying a policy, so another engineer told us we needed to shut the VM down and increase the RAM to 32GB, which I did today. 

However, after it rebooted, the VM showed 32GB in the summary page, and the UI of the management VM reflected that as well... but the physical mapped line in the VM performance pane and the VM listing pages at the host cluster and vCenter levels showed about 12GB, and over the past hour, the graph for it seems to have a very slight incline as time moves on, gradually moving from 11,189,xxx KB to 12,408,xxx KB as I move the mouse over the graph in performance monitoring of the Memory of this VM in vSphere.  Guest and Consumed memory (same line, different colors) is still only about 6.5% of that, at just about 2,xxx,xxx KB.  Why doesn't it show the 32GB?  It showed the 16GB line as mapped before (it was a flat line).  Am I reading things incorrectly?  Is this normal?

Labels (4)
0 Kudos
1 Reply
vbondzio
VMware Employee
VMware Employee

That really depends on how precise you are with the terminology and whether you are looking at guest or vSphere level counters. Is "Usage" memory usage of the VM object? That is active / touched memory and not really related to what a guest might need WRT resident memory to perform optimally. The reason "consumed" is slowly going up, is that with memory churn in the guest OS and it slowly touching the guest visible "physical" range and ESXi is backing that with large pages, i.e. your consumed memory will keep growing until it reaches the VM assigned memory, consumed will only go down when there is contention / you dip into the host clear / soft etc. states. I do explain consumed / active here to some extend: https://communities.vmware.com/t5/ESXi-Discussions/Reported-Memory-usage-on-SQL-VMs/m-p/2876282/high...

0 Kudos