Hi, I have a cluster of about 20 hosts and 400 VMs.
Vcenter 6.5 U2a, hosts 6.5
DRS is fully automated with default threshold level.
DRS keeps migrating same 5 VMs hundreds time per day.
Is there a way to understand why is doing that and why is choosing always same VMs?
Thanks in advance
I've seen this in conjunction with DRS rules. For example, with a rule that pinned some VMs to a few hosts and these hosts were heavily loaded. Then it moves the VMs that are least used back and forth all the time.
unfortunately this is not my case, all hosts are under 50% of utilization for both CPU and memory.
I'm wondering if is there a way to view inside logs why those vms are choosen?
How much RAM/CPU does those VMs have? Are those machines using memory/cpu intensive?
are all small vms with 2/3 Gb of ram with cpu and ram in idle most of the time
were you able to rectify it was showing such behavior?
If you found answer please let me know.
Do you have any DRS affinity rules or anti affinity rules?
not yet, currently we planned to upgrade vcenter to 6.5 2d cause in release notes we've found this:
"vSphere DRS might run migrations of virtual machines even when a cluster seems balanced
In vSphere 6.5, if vSphere DRS detects in a balanced cluster an imbalance between two hosts larger than a set migration threshold, it might run some migration operations that do not effectively improve the balance between the two hosts.
This issue is resolved in this release"
yes we have both
There's a fling for that!
It is called DRS Dump Insight: VMware Labs - Flings
it will help you figure out why VMs were migrated etc.