InstantRecovery to VMWare not working (Case-ID #04038486)

[Harald]

Hello,

I'm playin' around with Veeam 10, and one feature I'm really exited about is doing an instant recovery of a physical system. So, I've setup a Linux VM, installed Veeam agent in it, and backed it up with my Veeam 10 server as an agent job. That's working fine, but the instant recovery fails at "Converting Linux VM". Looking in the logs, I've found that Veeam is trying to connect to the IPv6 address of the VMWare server I've chosen for the recovery, and the task fails because this "network is unreachable".
The ESX-server has a IPv6 address in DNS, but it also has a IPv4 address, and I've already disabled IPv6 completely in the OS of the B&R server by setting the appropriate registry key. Veeam support suggested to add the vCenter by its IPv4 address instead of the name, but that didn't help. Backup is still working, the vCenter is reachable, but the conversion fails.
Any ideas about this? How can I configure Veeam to prefer the IPv4 address, even if it gets a v6 address from DNS?

Many thanks for any suggestion and best regards from Germany,

Harald

[Dima P.]

Hello Harald,

I've already disabled IPv6 completely in the OS of the B&R server by setting the appropriate registry key. Veeam support suggested to add the vCenter by its IPv4 address instead of the name, but that didn't help.

Can you please clarify if all the Veeam B&R components (i.e. repository, proxy servers and others) are connected over IPv4 as well? Can you please share the error code from the instant recovery log? Thank you!

[Harald]

Hello Dima,

thanks for your answer, and please excuse my late response - we're a bit busy with the actual Corona crisis.
I have only one server in my current setup which is working as proxy, Veeam- and repository server, so, the answer to your first question is "yes", there are no IPv6 connections between the parts since they're all running on one system.
Nevertheless, it seems that Veeam still tries to connect the ESXi-Server by its v6 address:

Code: Select all

"[VIX] Cannot invoke command 'cat /var/log/lvm.log'
[13.03.2020 19:41:25] <01> Error    Unable to connect to the remote server (System.Net.WebException)
[13.03.2020 19:41:25] <01> Error       at System.Net.HttpWebRequest.GetResponse()
[13.03.2020 19:41:25] <01> Error       at Veeam.Backup.ViSoap.VmGuestImpl.DownloadStrings(String fileName)
[13.03.2020 19:41:25] <01> Error       at Veeam.Backup.ViSoap.CSoapConnGuest.Popen(String command, String[]& stdout, String[]& stderr, TimeSpan timeout)
[13.03.2020 19:41:25] <01> Error       at Veeam.Backup.Core.CVixApplianceInvoker.InvokeCommand(String command, Boolean logInOut, TimeSpan timeout, String& output)
[13.03.2020 19:41:25] <01> Error    A socket operation was attempted to an unreachable network [2001:67c:510:96::1094:1]:443 (System.Net.Sockets.SocketException)
[13.03.2020 19:41:25] <01> Error       at System.Net.Sockets.Socket.DoConnect(EndPoint endPointSnapshot, SocketAddress socketAddress)
[13.03.2020 19:41:25] <01> Error       at System.Net.ServicePoint.ConnectSocketInternal(Boolean connectFailure, Socket s4, Socket s6, Socket& socket, IPAddress& address, ConnectSocketState state, IAsyncResult asyncResult, Exception& exception)
[13.03.2020 19:41:25] <01> Info     [Veeam.Backup.ViSoap] execute in shell [cat /var/log/lvm_error.log]"

Thanks for your answer and best regards from Germany,

Harald

[PetrM]

Hello Harald,

The stack trace above just says that we couldn't get a response to the web request but does not show where IPv6 came from.
I'd recommend to continue the investigation with our support team, apparently deep logs analysis is required.

Just one more idea:

How can I configure Veeam to prefer the IPv4 address, even if it gets a v6 address from DNS?

What about to add the corresponding record into hosts file on the Veeam server to resolve DNS name into IPv4 and to retry the test?

Thanks!

[Harald]

Hello PetrM,

yes, I was also thinking of the hosts file. So, I added the IPv4 adress to it, but, Veeam still tried the IPv6 address, ignoring my hosts entry. I'm already working on this with the support team, they made a couple of good suggestions already, but we've unfortunately not found a real solution.

Thanks for your reply and best regards from Germany,

Harald

[PetrM]

Hi Harald,

Do not hesitate to request escalation of the case.
I think the main question now is to detect the source of this IP or to explain why hosts file does not work if this IP comes from DNS.

By the way, it could be also VCenter server and not ESXi: "[2001:67c:510:96::1094:1]:443" (443 port is used for connection).

Thanks!

[Harald]

Hello Petr,

thanks again for your answer, you're right - the main question is why Veeam still uses the IPv6 address (which is indeed the ESXi address, and not the vCenter - I've already checked this).
And, escalating is an option, I've just uploaded a bunch of new logfiles and will give support team some time to analyze them, hopefully they can figure out something.

Thanks and best regards from Germany,

Harald

[Harald]

Hello all,

and apologies for the late reply: I finally got it working with Veeam support, the cause is that our ESXi-Servers were registered in the vCenter with their IPv6 addresses.
It seems that Veeam asks the vCenter and not the DNS or the local hosts file to get the destination address, and since the vCenter replies with an IPv6 address Veeam tries to connect to it - no matter if IPv6 is enabled in the OS or not.
I've tested it by adding one of the ESXi servers as a standalone server to Veeam itself with its IPv4 address, and the restore worked like a charme. Good to know for the future, so, if anyone uses IPv6 with vSphere be sure to use the IPv4 addresses in the vCenter.

Thanks to all who replied, and thanks to Veeam for the good support,

Harald