Forum Discussion

mmeeggaann's avatar
mmeeggaann
Level 5
13 years ago

Use dedicated backup network to back up cluster resources

Using Backup Exec 2012 R3, we have to back up  3 HA clusters, each one with 2 servers:

Exchange 2010 DAG

Microsoft Cluster for SQL Server 2008

Hyper-V HA

 

We are trying to use a dedicated backup network and the backup jobs are configured to do so (Network and security settings). All the backups of any high available resource are failing with the following error:

 

The job failed with the following error: The media server could not connect to the remote computer. The remote computer may not exist on the selected subnet. Try an alternative network interface, or allow Backup Exec to choose any available network interface.
 
 
I found at SGMon log that BE queried its database and found a record for the  virtual cluster hostname associted to a  Public (or productive) IP address, then it's trying to connect to it from the  backup network interface:
 
SQL backup:
 

 

BENGINE:  [07/17/12 19:55:31] [6592]     [server]             - Checking for Evaluation expiration.

BENGINE:  [07/17/12 19:55:31] [6592]     [engapi]             - ENGSCRPT: ConvertDeviceName osId = 0Xe

BENGINE:  [07/17/12 19:55:31] [6592]     [engapi]             - ENGSCRPT: IN - oldDeviceName = \\MSSQLSERVER2008.domain.com\H:

BENGINE:  [07/17/12 19:55:31] [6592]     [engapi]             - ENGSCRPT: OUT - newName = \\MSSQLSERVER2008.domain.com\H:

BENGINE:  [07/17/12 19:55:31] [6592]     [engapi]             - SetupCryptInfoInBSD: Backup encryption options: bHWEncryption=0,bSWEncryption=0,bSWEncryptionIfNoHW=0

BENGINE:  [07/17/12 19:55:31] [6592]     [engapi]             - ENGSCRPT: No script found

BENGINE:  [07/17/12 19:55:31] [6592]     [server]             - Failed to Load AllowExcludes key, We will update DR file evenif there are userExcludes. Error: 5

BENGINE:  [07/17/12 19:55:31] [6592]     [loops]              - NDMPAgentConnector:SetupConnection(): Data Server = MSSQLSERVER2008.domain.com

BENGINE:  [07/17/12 19:55:31] [6592]     [ndmp\ndmpcomm]      - Could not resolve the "bews-ndmp" or the "ndmp" service, error code: 10109, using port 10000

BENGINE:  [07/17/12 19:55:31] [6592]     [ndmp\ndmpcomm]      - ndmpConnectEx: Querying the neighbour advertisement cache to discover information on 'MSSQLSERVER2008.domain.com' ...

BENETNS:  [07/17/12 19:55:31] [8308]     NRDS API - client connected.

BENETNS:  [07/17/12 19:55:31] [8308]     Connecting to BE Database.

BESERVER: [07/17/12 19:55:31] [8548]     -1 Client requested key (1342573096).

BESERVER: [07/17/12 19:55:31] [8548]     01 Server Configuration: Client added: 9

BESERVER: [07/17/12 19:55:31] [8548]     -1 Client 'MEDIASERVER' connected('','DOMAIN\LOGONACCOUNT'): 0xb586e30

BENETNS:  [07/17/12 19:55:31] [8308]     Successfully connected to BE Database.

BENETNS:  [07/17/12 19:55:31] [8308]     Reading agent database record for MSSQLSERVER2008.domain.com.

BENETNS:  [07/17/12 19:55:31] [8308]     Found agent record 22 for MSSQLSERVER2008.domain.com.

BESERVER: [07/17/12 19:55:31] [8548]     01 Server Configuration: Client removed: 8

BESERVER: [07/17/12 19:55:31] [8548]     -1 Client 'MEDIASERVER' Disconnected:0xb586e30

BENETNS:  [07/17/12 19:55:31] [8308]     Disconnected from BE Database.

BENETNS:  [07/17/12 19:55:31] [8308]     NRDS API - client disconnected.

BENGINE:  [07/17/12 19:55:31] [6592]     [ndmp\ndmpcomm]      - ndmpEstablishConnectionUsingGivenSubnet: '10.10.2.231' and '1.1.1.32' were not found to be on the same subnet

BENGINE:  [07/17/12 19:55:31] [6592]     [ndmp\ndmpcomm]      - Could not connect to any of the remote addresses using subnet: '10.10.2.224' and mask: '255.255.255.224'

BENGINE:  [07/17/12 19:55:31] [6592]     [ndmp\ndmpcomm]      - ndmpEstablishConnectionUsingNetworkOptions: Could not connect to the remote host through NIC: 'Bakcup'

BENGINE:  [07/17/12 19:55:31] [6592]     [ndmp\ndmpcomm]      - ndmpConnectEx: unable to connect using NetworkOptions to MSSQLSERVER2008.domain.com

BENGINE:  [07/17/12 19:55:31] [6592]     [ndmp\ndmpcomm]      - ndmpConnectEx: no NIC fallback to MSSQLSERVER2008.domain.com

BENGINE:  [07/17/12 19:55:31] [6592]     [loops]              - NDMPAgentConnector::Connect: ndmpConnectEx() failed on server MSSQLSERVER2008.domain.com, port: 0.

BENGINE:  [07/17/12 19:55:31] [6592]     [loops]              - NDMPAgentConnector::HandleConnectionError: ndmpConnect failed: The remote machine MSSQLSERVER2008.domain.com was not detected on the selected subnet 10.10.2.224/27 via network interface Bakcup, and the 'Fallback' option was not chosen.

BENGINE:  [07/17/12 19:55:31] [6592]     [loops]              - IsLocalAgent:beclass::BEConvertServerNamePlus returned error: 87

BENGINE:  [07/17/12 19:55:31] [6592]     [loops]              - NDMPRAHostPreScanEngine::ConnectToDataServer() failed. Error = 0xE0000F02.

BENGINE:  [07/17/12 19:55:31] [6592]     [loops]              - NDMPHostPreScanEngine:ConnectToDataServer() failed for machine MSSQLSERVER2008.domain.com. Skipping BSDs for this machine.

BENGINE:  [07/17/12 19:55:31] [6592]     [server]             - MountMultiple: Created MountThread thread 8800

 
 I already added the entries in the media server hosts file and the remote servers as well. However it seems as Backup Exec has registered in its databases the wrong IP address, how do I clean those entries in BE databae?
 
I also modified the publishing settings at the remote agents.
 
 
Is there something else that must be done to be able to back up high available data through an isolated network?
 
  • Does the Virtual Server (in the cluster config) for MS SQL have an IP Address on both NIC/subnets?

    Also I hope you are not trying to use Hyper-V Agent for the backups as we do not supported distributed enviroments - which would mean Clustered SQL or Exchange DAG would not be supported for a Hyper-V with GRt enabled backup.

  • You got to make sure that your dedicated backup network is set up correctly.  I presume that you have two NIC's,  the normal one and the one to the backup network.  Unplug the normal NIC and then ping the media server from the remote server and vice-versa.  If the ping test fails, then you have not set up the backup network correctly.  You might need to update the Hosts files on the remote and media servers to include the network address of the servers using the sub-net of the backup network.  Also, make sure that the remote agent is publishing to the media server using the backup network.

  • in the selection list, have u define the backup ip of the machines being backup using the user-defined selections? normally this is how i forced the backup exec server to use the ip.

  • Does the Virtual Server (in the cluster config) for MS SQL have an IP Address on both NIC/subnets?

    Also I hope you are not trying to use Hyper-V Agent for the backups as we do not supported distributed enviroments - which would mean Clustered SQL or Exchange DAG would not be supported for a Hyper-V with GRt enabled backup.

  •  

    Thanks for your quick responses. 
     
    pkh, the backup network is working well. I'm able to back up local resources of each server through backup network. The problem comes out when I select cluster resources. I did some test with the servers that are part of the SQL cluser... I set the remote agent to publish to  media server through the backup network IP address... and I also added  in the hosts file media server  the  backup Ip addresses of each node of the cluster. The problem still happens.
     
    oithim_yew I have not tried  for windows servers the alternative you mentioned... I'll test it.
     
    Colin, In this particular case it won't apply ... the Exchange DAG and SQL clusters are physical servers. I also think that under cluster config must be an IP address for the backup network and that's what I suggested to the cluster admin, because now the configuration is not in that way. I'm waiting for the change is applied.
     
    So... what about the DAG or the Hyper-V HA... do you have an idea how must be configured an isolated backup network?
     
     

     

  • Colin, if I understand well... there must be an virtual IP address of the backup network associted to  the cluster hostname as exists for the production network?

     

  • After the backup network was addded to the cluster configuration as a resource, the jobs complete succesfully. 

     

    Regards!!