Kanstantsin
Enthusiast
Enthusiast

My lub:

vCenter 6.0

SRM 6.0

IBM SRA - IBMSVCSRA_2.3.1.0 for IBM v7000

There is VM on protected site. When I disconnect storage from host and do "run recovery plan(desaster recovery)" SRM start VM on recovery site and don't shutdown VM on protected site.

Eventually I have two VMs in power on(on protected site and on recovery site). VM on protected site respond on ping and have opened ports(example 3389 - RDP) but I cannot connect to RDP and console to this VM.

Is this normal?

And second question, How VM without datastore can work?

0 Kudos
1 Solution

Accepted Solutions
vbrowncoat
VMware Employee
VMware Employee

This is normal behavior. When you run a recovery plan in DR mode it will attempt to shutdown VMs at the protected site but if it is unable to will just continue with recovery plan.

What you are experiencing with the VM with disconnected storage is kind of a "zombie" VM. When you disconnect storage the VM often continues to run because there are a number of things stored in memory that allow it to function in a limited sense. As you experienced it may even be able to respond to pings. More than likely though it is really unable to do any work and therefore "dead" but not aware of it. Until vSphere 6 there was no simple way to deal with this. Now in vSphere 6 there is VMCP (VM Component Protection) which will monitor connectivity to VM storage and if an issue arises (like an APD or PDL) VMCP will shutdown the VM and restart it on a healthy host (eg. one that has access to storage).

Does this answer your questions?

View solution in original post

0 Kudos
2 Replies
vbrowncoat
VMware Employee
VMware Employee

This is normal behavior. When you run a recovery plan in DR mode it will attempt to shutdown VMs at the protected site but if it is unable to will just continue with recovery plan.

What you are experiencing with the VM with disconnected storage is kind of a "zombie" VM. When you disconnect storage the VM often continues to run because there are a number of things stored in memory that allow it to function in a limited sense. As you experienced it may even be able to respond to pings. More than likely though it is really unable to do any work and therefore "dead" but not aware of it. Until vSphere 6 there was no simple way to deal with this. Now in vSphere 6 there is VMCP (VM Component Protection) which will monitor connectivity to VM storage and if an issue arises (like an APD or PDL) VMCP will shutdown the VM and restart it on a healthy host (eg. one that has access to storage).

Does this answer your questions?

View solution in original post

0 Kudos
Kanstantsin
Enthusiast
Enthusiast

Yes, thank you

0 Kudos