I'm running VAL on an Ubuntu 14.04 server to back up to a VBR server. It keeps trying all the available IP addresses on the VBR server, despite being configured with the specific IP it should use to connect to it. This slows down the start and running of each job by qiote a bit. Is there any way to force it to only use the one IP?
When VBR server is contacted by Agent it returns all IP addresses of the host where the target data mover is working, i.e. a repository or its gateway. In case the host has multiple IP addresses then multiple IPs will be returned.
Currently there is no conventional fix for that, however we're considering to fix this in future release. It should be possible to workaround the issue by setting an iptables rule that will forwards all port 2501 outbound traffic to the correct IP. Please note, that if there are multiple agents sending backups to the same repository then ports next to 2501 will be assigned for data transfer, e.g. agent1: 2501, agent2: 2502, agent3:2503.
I have exactly the same issue and I believe that is a bug in Linux agent and I made the following workaround to avoid communication between Linux Agent and Veeam Backup Server:
1 - Created a shared folder into backup repository (CIFS or NFS)
2 - Mount the shared folder in Linux
3 - Configure the backup job from the Veeam Agent for Linux utility instead of using Veeam Backup Server console.
In this way, the backup will be managed by the agent instead of Backup Server and the backup files will be saved directly on shared folder into backup repository.
Before this, I did try to create iptables NAT rules but doesn't worked
Anyway, will be useful Veeam Support check this case, because when the Backup Server has multiple IPs, the Linux agent can face issues.
Anyway, will be useful Veeam Support check this case, because when the Backup Server has multiple IPs, the Linux agent can face issues.
On Agent - did you specify an IP address or DNS name of VBR server?
Where does your VBR repository reside - is it on VBR server or on a separate machine? If the latter, does the machine have multiple IP addresses?
Then it is exactly the same behaviour that I've described earlier - if target repository has multiple IP addresses, then VBR will return all of those addresses to VAL. In your case the storage is connected to VBR via iSCSI therefore data-mover service runs on the very same machine with VBR, therefore it returns all available addresses to VAL. The same behaviour can be observed if you try to replace VAL with VAW.