Slow "undeploy" process


Because of in my scenario it is necessary to instantiate VMs with size 50 GB (yeah… ufff), I’m forcing these VMs to be instantiate in my second server (more CPUs). My datastores are TM_MAD=ssh (each of my two servers has its own datastore), so instantiation time is loooooong (network 1 Gb). This problem will be solved (I hope) when I insert a PCI-Express network card 10 Gb to connect two servers…

Now, I have two question:

  1. If I instantiate these VMs in my second server (not scheduler server) and, then, I undeploy that VMs, I have noticed that VMs “returns” to main server.That process, multiplied by “n” VMs, takes a loooot of time but, why “undeploying process” needs to return VM to main server? Can’t VM stay “undeployed” at second server if, also, I have scheduled them forcing “scheduling in second Host”? What is the reason to transfer VM to main server?

  2. In these VMs, yesterday, I applied some “action”. First of them is “run a resume” at 8:30 AM and, then, I also applied three more “action” to run “undeploy” at 9:30 AM in three specific days. During night, at 03:00 AM, I run my undeploy script to run an “undeploy” to all running VMs. However, this night, my script has been run OK… but when VMs have been undeployed, 25 seconds later script to “resume” has been run… Hour: 03:50:26 (undeploy has taken 50 minutes… because, I suppose, 50 GB size VM and transfer to main server). Why has been executed “resume”? Hour was 03:50:00 AM, not 8:30 AM.


Take a look at OneStor datastore. The image caching mechanism should be what you need.

1 Like


I will have a look in my testing environment.