Forum Discussion

SYMAJ's avatar
SYMAJ
Level 6
11 years ago
Solved

VM Restore with no vCenter Server

I have an environment whereby there is a single vCenter server managing ESX clusters spread accross two locations.  I am currently testing a restore of a VM in the remote location when I have no access to the vCenter server (for DR purposes).

I have defined the ESX servers in the remote site within the credentials section of NBU (7602) as ESX Restore Servers, and am attempting a FULL VM restore specifying NONE and the vCenter server and a specific ESX server as the alternate location.  The credentials being used when I defined the ESX Restore server are root.

The restore is failing (as per joblog below) when trying to create the VM on the ESX server - and I believe this is related to the ESX server being currently managed by vCenter.  Also, I believe this is a 'problem' introduced with ESX 5.x.  I don't know why the joblog states it is restoring to vCenter Server ESX2.xxxx.com - this is an ESX server not a vCenetr server.

In a DR scenario, when there is no access to vCenter, I see the options being either recreate the vCenter server or amend the ESX servers so as they are NOT MANAGED by vCenter.  However, for DR testing purposes I don't want to do either of these options - is there another way of successfully restoring a VM directly to an ESX Server ?

There are similar references to this in TECH108679, where it also states:

VMware does not support restoring a virtual machines directly to an ESX 5.x server that is managed by vCenter. If the ESX host is managed by a virtual center, it is necessary to select the virtual center during the restore process.

However if the ESX restore credential exist for the ESX host, NetBackup will only use the virtual center to create the VM and it will use the ESX host for the writing operation.

This is covered in the NetBackup 7.5 VMware Administration Guide on Page 41:
"For large environments with hundreds of hosts, NetBackup may not need full access to the vCenter server. With the restore ESX server type, you can give NetBackup full access to a single ESX server that is dedicated to restore."

Thanks,

AJ.

11:41:18 21/05/2014: Restore Started

11:41:22 (878.001) Restoring from copy 1 of image created 21/05/2014 11:00:21 from policy VMware-Backup
11:41:22 (878.xxx) Restoring to vCenter server ESX2.xxxx.com, ESX host ESX2.xxxx.com, Datacenter None, Folder None, Display Name DG-V-EVDA01_RESTORED, Resource Pool/vApp /ha-datacenter/host/ESX2.xxxx.com/Resources, Datastore/Datastore Cluster EQL-Datastore-3
11:41:25 INF - Beginning read-blockmap on server nbu-76.xxxx.com of client dg-v-evda01, reading file @aaaab.
11:41:25 (878.xxx) Executing Virtual machine creation on nbu-76.xxxx.com to server ESX2.xxxx.com.
11:41:36 (878.001) Status of restore from copy 1 of image created 21/05/2014 11:00:21 = Restore failed due to failure to create the virtual machine

11:41:36 (878.001) The following files/folders were not restored:
11:41:36 (878.001) UTF - /EQL-Datastore-1/DG-V-EVDA01/DG-V-EVDA01.vmdk
11:41:36 (878.001) UTF - /EQL-Datastore-2/DG-V-EVDA01/DG-V-EVDA01.vmdk

11:41:36 (878.xxx) INF - Status = VMware policy restore error.

 

  • I believe that whilst an ESX server is being managed by vCenter you cannot directly restore to it as the creation of the machine on the ESX node needs to be done from vCenter.

    I convonced the customer to create a second vCenter server to manage the ESX nodes in the DR site and all is now well.

    Thanks for the input.

    AJ

6 Replies

Replies have been turned off for this discussion
  • Additional error log info:

    21/05/2014 11:41:25 - begin reading

    21/05/2014 11:41:25 - end reading; read time: 0:00:00

    21/05/2014 11:41:26 - Info bpdm(pid=3220) completed reading backup image        

    21/05/2014 11:41:26 - Info bpdm(pid=3220) EXITING with status 0        

    21/05/2014 11:41:26 - Info nbu-76.xxxx.com(pid=3220) StorageServer=PureDisk:nbu-76.xxxx.com; Report=PDDO Stats for (nbu-76.xxxx.com): read: 10804 KB, CR received: 26695 KB, CR received over FC: 0 KB, dedup: 0.0%

    21/05/2014 11:41:36 - Info bpVMutil(pid=4264) INF - vmwareLogger: WaitForTaskCompleteEx: Access to resource settings on the host is restricted to the server '192.168.10.128' which is managing it. <239>

    21/05/2014 11:41:36 - Info bpVMutil(pid=4264) INF - vmwareLogger: WaitForTaskCompleteEx: SYM_VMC_ERROR:  TASK_REACHED_ERROR_STATE     

    21/05/2014 11:41:36 - Info bpVMutil(pid=4264) INF - vmwareLogger: CreateVm: SYM_VMC_ERROR:  TASK_REACHED_ERROR_STATE     

    21/05/2014 11:41:36 - Info bpVMutil(pid=4264) INF - vmwareLogger: CreateVirtualMachineExAPI: SYM_VMC_ERROR:  TASK_REACHED_ERROR_STATE     

    21/05/2014 11:41:36 - end Restore; elapsed time: 0:00:16

    VMware policy restore error(2820)

  • symaj, think you need to remove the ESX host in question from the vcenter before attempting the restore directly to the ESX host. Which simulate the disaster scenario you want to test.

     

  • AJ

     

    Why don't you look into using instant recovery - it looks like you are running 7.6.0.x and using a disk pool, so this might be a better option for the initial recovery. That way you can spion up the server before vmotioning it off to the appropriate ESX server.

    However, I think you may still run into a similar issue with attempting to control the ESX host directly while it is still "attached" to the vcenter server as per the post above.

     

    Cheers
    David

  • Dosn't instant recovery require vcenter access to work ? Even if it dosn't, storage migration in vmware does

  •  

    make the policy query with  select any esx server and not vcenter.

    The backup transfer will going via ESX server  and  recovery can initiated with the vcenter , recovery transfer is going in the same way direct back to the esx server.

  • I believe that whilst an ESX server is being managed by vCenter you cannot directly restore to it as the creation of the machine on the ESX node needs to be done from vCenter.

    I convonced the customer to create a second vCenter server to manage the ESX nodes in the DR site and all is now well.

    Thanks for the input.

    AJ