Netbackup 7.0 failing on W2K8 Exchange 2007 (x64) Clusters W/ Multiple NICs and a Private Backup network

[Tacfire]

Morning Experts,

I'm having a serious issue trying to get Netbackup 7.0 configured to run properly on a Windows 2008 / Exchange 2007 CCR Cluster environment.  BTW, every Server has Eight (8) Network Cards, just to add alittle complexity to an already complex product.   When I try to perform a VSS Exchange snapshot backup, the job fails multiple times, but will randomly complete successfully.  It appears that Network may have something to do with this, but I have Host files deployed on each Media Server, and the Master Server.  The Master Server Host File contains entries for all Media Servers (Server Name is entered identical as it appears in Production and DNS, but with the Backup Network Interface).  The rest of the Media Servers contain a Host File with only the Master Server, against actual name, and IP for the Backup network.  When I PING the Master Server from any Media Server, it answers with the correct Backup IP, and the same applies for the Master PINGing the Media Servers.  Does anyone have any ideas?

 and I am getting constant Errors in the Logs;

Error 25
Error 103
Error 130

Here's a basic summary of my environment:

1 AD Domain

1 Exchange Organization

Netbackup Setup:

1 Master
18 media Servers

My environment:

18 Windows 2008 (x64) Clusters running Exchange 2007 (SP2, Rollup 4) CCR (Each Cluster node is configured with Eight (8) Network Cards.  Here's the breakdown - 2 Node Clusters), Every Server is configued identically; I have Two (2) NICs teamed for Production, Two (2) Teamed for CCR Traffic, One (1) Backup Network, One (1) Management Network, and One (1) Security Network, and of course One (1) Heartbeat for the Clusters.  Each Cluster Node is a Media Server 

1 Master Server, which is configued with Two NICs, One (1) Backup, and One (1) Management.  Note that they Backup Master Server is prohibited from talking over the production network for security reasons, and a requirement of my customer.  


 

[Jonathan_Bourke]

In my experience, name resolution issues are 99% of NetBackup issues. Solid forward and reverse name lookups really really help.

That said, I have been there with Exchange 2007 CCR and a backup LAN, although the exchange servers were only dual homed. Your big issue here is indeed name resolution and AD. Exchange CCR clusters rely on AD for naming, etc. whereas NetBackup doesn't give a damn about AD. Exchange will use the short NETBIOS name or HOSTNAME for individual nodes within the cluster, and NetBackup will typically either not be able to find this names, or not be able to route to them.

Assuming that each client is named HOST, and exchange CCR is CLUSTER, and backup network is either HOST_BKUP or HOST.BKUP, then my suggestions would be:

1) Install and configure NetBackup client as HOST_BKUP
2) If possible configure a bkup network as a Cluster Resource, CLUSTER_BKUP
3) Add clients HOST_BKUP to a file system policy and CLUSTER_BKUP to an exchange policy
4) Now, add aliases for all other possible hostnames or domains to each client to the EMM database as follows: 
        
       nbemmcmd -addhost -machinename HOST_BKUP -machinetype client -operatingsystem windows -netbackupversion 7.0 -masterserver MASTER
       nbemmcmd -machinealias -addalias -alias HOST -machinename HOST_BKUP -machinetype client
       nbemmcmd -machinealias -addalias -alias HOST_CCR -machinename HOST_BKUP -machinetype client

5) Disable reverse DNS lookups on all clients, as contacting a CLUSTER name and then getting a NODE name will really make that fun.
6) Enable logging on your all your clients and servers, particularly BPCD in this case. 

This should hopefully get you on your way. 

Just wait till you get to Exchange 2010 DAG's passive node backups on multihomed servers, good fun!

Jonathan