Backup a NLB Cluster / dual homed systems
Hi
We are trying for the first time to backup a NLB Cluster. I didn't find any Information about this.
The backup always fails with "V-79-57344-65072 - The connection to target system has been lost. Backup set canceled".
We had this error before with dual homed server. Took us a while to find out there is an option to define the NIC which the backup uses.
But the backup job is set to use a specific NIC, the two server in that NLB both got that NIC configuration.
The backup is set up to use the internal physical node IP to backup those two server.
Any documents flying around for NLB backup? Or any other Idea what's wrong?
NLB Cluster IP: 172.20.32.0 /24
Physical Node IPs internal: 172.20.0.0/19
Physical Node IPs external:172.20.32.0/24
Backup Interfaces : 172.20.0.0 /19
Backup Exec Server IP: 172.20.1.133/19
Now as soon as I don't select the cluster systems via name, but put there IP into the selection list the backup seems to work.
If I select them with their names the backup seems to run over the Firewall and connects to the 172.20.32.0/24 Interfaces.
I assume that actually this is the problem, but I don't get it why backup exec runs over that Network.
The job has the setting to use the 172.20.0.0/19 Interfaces. For testint purpose I also didn't select the Option to allow an other NIC. But it still uses the wrong one. The Media Server itself doesn't even have a 172.20.32.0/24 Interface. So it takes the route:
Backup Server 172.20.0.0/19
Firewall
DMZ Network
NLB Cluster Systems 172.20.32.0/24
A normal ping from the backup server to a physical node of the NLB-Cluster goes trough 172.20.0.0/19.
We had the same error with all the other normal dual homed server. Till we defined a specific NIC for the backup to force it over the internal Network.
But for a reason I don't get till now, the NLB-Cluster nodes backup seems to ignore this.
Of course I could select the systems from the selection list via IP, but that seems more like a work around which shouldn't be needed.
Comments
check the following 1.
check the following
1. install remote agent on both servers (server 1 & server 2)
2. make sure that remote agent service is set to automatically and started
3. ping on virtual (external) ip
5. telnet to all IPs (external) onn port no. 10000, 445
6. add backup server entry in both servers host file
please check and confirm.
Please check and confirm. If this remedy worked for you Kindly mark it as a SOLUTION
I can do that but to be
I can do that but to be clear:
To backup shouldn't go over the external IP, which it tries to do if I select the systems by name.
I always thought setting the Job to use an specific interface should also force the use of that interface on the system you backup.
Running the backup over the Firewall causes tremendious traffic, leading to Problems for all the other systems traffic going over the firewall.
So the primary problem here is:
It uses the external Intefraces.
It shouled use the internal.
It works fine if I select the systems I want to backup via IP and not by Name.
So did I missunderstand to function to set the NIC inside the backup job?
so when you ping to virtual
so when you ping to virtual name from backup server, which IP address it is showing?
Please check and confirm. If this remedy worked for you Kindly mark it as a SOLUTION
it get's both the internal
it get's both the internal and the external IP with nslookup if I ping the physical members.EDIT: Noticed that nslookup only answers with internal and external interface for one server. We changed that so now all NLB Members only deliver there internal IP via DNS. But Backup Exec still goes over the external.
What do you mean by virtual IP? The Cluster IP? there only the external exists. The internal Interface is not clustered.
ping runs over IPv6 internal IP.
Still shoulen'd Backup Exec take the internal way if I configured the NIC it sould use?
The backup is supposed to use the internal IPv4 172.20.0.0 /19.
The Backupserver itself only has that Interface, but the backup runs over the Firewall into the DMZ.
There it's running over 172.20.32.0/24 Even thought there would be a direct route trough the 172.20.0.0/19 Network.
I always thought that's the Idea behind this setting. And on all other server the problem dissapeared with that setting.
If I ping from the NLB members to the backup server the connection runs over IPv6. Nslookup delivers only the internal IPv4 Adress.
Someone can tell us at least
Someone can tell us at least how backup exec even gets the DMZ IP?
ARP -A doesn't deliver it.
Nslookup also doesn't deliver it.
Both are delivering the intern IP, but backup exec decides to go the other way trough the firewall into the DMZ.
As I sad we could select the system in the select list with the internal IP and not over the name, but that would be an exception which means we need to inform a lot of people about this and document it.
Also why are these two server the only one that got this problem even though there are other server inside the DMZ backed up with the same job?
Would you like to reply?
Login or Register to post your comment.