Your Browser is not longer supported

Please use Google Chrome, Mozilla Firefox or Microsoft Edge to view the page correctly
Loading...

{{viewport.spaceProperty.prod}}

Restoring VM definitions after an outage

&pagelevel(2)&pagelevel

Managing VM definitions

A VM definition that is also unambiguous in the SU Cluster exists for every BS2000 VM.
The VM definition contains a description of attributes as well as resources and devices, which are assigned to the BS2000 VM or should be assigned during their activation. A VM definition contains also the current state of the BS2000 VM.
All VM definitions are managed on the CRD (see chapter -> CRD functionality; CRD with external configuration disks).

The VM definitions are identified unambiguously by their VM name and the SU to which they are assigned, and a BS2000 VM can run only on such that SU to which its VM definition is assigned.
This is why the following is valid: to start a BS2000 VM on a specific SU of the SU Cluster, its VM definition must be assigned to the desired SU.

Command /RECOVER-VM-DEFINITION

If an SU to which there are VM definitions assigned is not available, then the VM2000 administrator can transfer the VM definitions with the command /RECOVER-VM-DEFINITION to the available SU of the cluster and start the VM there.
This situation can occur for example during deactivation caused by maintenance of an SE Server or an SU or after an SU outage or server outage.
This means that the command /RECOVER-VM-DEFINITION supports the customer’s HA procedures and processes (HA means High Availability in this case) to make a quick restart of the productive BS2000 systems on another SU after an outage (failover).

The command /RECOVER-VM-DEFINITION is available in any SU Cluster (SU /390 and SU x86) with VM2000. The (apparent) status of the VM displayed by SEM does not play any role for this command, but it must be ensured that the remote SU is not running.

Example:

The VM definition of the guest system with the VM name VM12PROD is taken from the SU ABSEE701 to the local SU.

/RECOVER-VM-DEFINITION VM-NAME=VM12PROD

In a SU Cluster with two SU the name of the removed SU is unambiguous and known to the VM2000, so it can be left out.
The name can still be provided though (both commands are equivalent):

/RECOVER-VM-DEFINITION VM-NAME=VM12PROD,SERVER-UNIT-NAME=ABGSE701

This command is only performed when the following conditions are met:

  • The local SU on which the command was entered and the remote SU specified in the command are part of the same SU Cluster.
  • The monitor system (SU /390) resp. the X2000 (SU x86) of the remote SU has not written any alive message on the CRD in the intended time.
  • The VM name is not used locally.
  • The max. number of VM definitions is not reached on the local SU.

After performing the command /RECOVER-VM-DEFINITION the VM definition is in state DEFINED_ONLY and the VM is not initialized, so it doesn’t allocate any resources.

The VM definition can either be put in operation immediately with the command /ACTIVATE-VM-DEFINITION and /START-VM or adjustment is still needed.

Possible scenario

A change of device assignment (/MODIFY-VM-DEFINITION) for example can be necessary if not only the SE Server but also the storage system with the source disks is damaged in a redundantly designed computer center (total outage of a site).

The system manager activates the mirror disks from the other SE Server of the SU Cluster in the other storage system (through StorMan in SE Manager of the MU or SHC-OSD in BS2000) and changes the MNs of the disk devices in the VM definition as shown in the below example.
Afterwards the VM definition is activated and the VM is initilized by this and finally the BS2000 is loaded from the mirrored IPL disk.

The reservation of the resources described in the VM definition is done with the command /ACTIVATE-VM-DEFINITION.
From this point in time, the VM attributes and device configuration can be modified with the usual VM2000 commands such as /MODIFY-VM-ATTRIBUTES or /ADD-VM-DEVICES. The changes are done then on the initialized VM and in the VM definition.

Example of changing the device configuration in the VM definition of the VM VM12PROD and the following start of the VM:

/MODIFY-VM-DEFINITION VM-NAME=VM12PROD,DEVICES=*REMOVE(UNITS=*DEVICERANGE(
    FROM=A000,TO=A0FF))
/MODIFY-VM-DEFINITION VM-NAME=VM12PROD,DEVICES=*ADD(UNITS=*DEVICE-RANGE(
    FROM=B000,TO=B007,USAGE=*EXCLUSIVE))
/MODIFY-VM-DEFINITION VM-NAME=VM12PROD,DEVICES=*ADD(UNITS=*DEVICE-RANGE(
    FROM=B008,TO=B0FF,USAGE=*SHARED))

These commands cause that devices B000 –B0FF are assigned during initialization of the VM instead of the previous devices A000 – A0FF.
The devices B000 – B007 are assigned exclusively to the VM, the devices B008 – B0FF are assigned as shared.

The command for displaying the devices contained in a VM definition is done with the command /SHOWM-VM-DEFINITION, so for the above example:

/SHOW-VM-DEFINITION VM-NAME=VM12PROD,INFORMATION=*UNITS

Finally, the VM definition can be activated and the VM can be started:

/ACTIVATE-VM-DEFINITION VM-NAME=VM12PROD
/START-VM IPL-UNIT=B000,VM-ID=VM12PROD