We recently started utilizing the resizing recommendations, setting it to "manual" mode so we can 1-click resize application servers during our maintenance windows.
Since a reboot is involved, it is disruptive and we don't want this happening live.
Something we noticed is that when the hypervisors (vcenters) are busy, they show up in the resize action and will show up with a check-box to resize.
Not only will this not work if you click it, it may take the vcenter down and require manual intervention to get it back up.
Two ideas here:
The resizing automation should either WORK on the hypervisor (it should know how to resize vcenter by going to the ESX host directly and resizing it there) or it should not show up in the task list.
The application can find the list of things to not allow EITHER manual or automatic resizing of in the "target configuration" section.
If a resizing opportunity matches a listed in the target configuration, don't allow manual resize action UNLESS THE APPLICATION KNOWS HOW TO DO IT.
Same goes with the SQL backends that may support your venter environment. Not sure how it can tell what these are, but it doesn't need to be resizing these manu-matically either.
Lastly, something we all like is the "mass-resizing" feature that is what the task list effectively is.
We see no other way to accomplish this with any other tool in vmware...meaning, this is a FEATURE that I don't even believe is accessible without a specific resizing recommendation being in the tasklist.
We created a monthly report that emails our jira mailhandler to open a ticket to process the resizing requests.
What would be AWESOME is if this list were available when the operator logs in, so they can action directly off it.
The only caveat is that you will need the ability to deselect certain resizing recommendations and not take them.
Such as, DR hosts that have no load normally but need to maintain vcpu and mem settings, and 32bit windows machines that cannot use more than 4GB of ram. Our system recommends 8GB for 32bit machines so we had to create a custom group, find these by name and then turn off manual resizing which is unfortunate because we still want the cpu recommendations here, just not the memory ones.
Considering a monthly or other scheduled resizing schedule, you need to know more about what you get on a daily/weekly/monthly report. For example, if I want to get resizing recommendations based on a prior week, what am I getting in the report? Is it a collection of everything that would have hit the daily report, or is it only the items that chronically need resizing over the report period?
This will tell us whether we need to customize this to see "top resizing opportunities" over a period, so we know we need to action all of them.
Then, to reduce repetitive ignores, we would like to mute resizing recommendations for hosts that hold capacity in reserve, like DR hosts and QA and build machines.