Solved

Configure AMI / timeout / retries for Veeam Proxy Appliance in AWS


Userlevel 2

Hello,

I’m trying to restore a Veeam backup from Capacity tier (S3) into an EC2 instance in AWS.

However, I’m facing a problem with the proxy instance. Once I start the restore process, the proxy instance is created, but after abou a minute I get an error:

“Preparing proxy appliance VM Error: The destination would not be a correct SSH server.”

I tried ssh to the proxy, but the connection gets rejected, while connections to other hosts in the same subnet with the same security groups are successful.

So I manually created an instance using the same settings Veeam uses for the proxy, same subnet, security groups, AMI, and found that connections are successful, but only a few minutes after instance creation with that particular AMI Veeam chooses for the proxy instance (aws-parallelcluster-3.1.4-ubuntu-2004-lts).

Now, the question is, is there a way to configure the AMI, the timeout or the number of SSH retries for the proxy appliance?

Thanks!!!

icon

Best answer by pampelix 24 May 2022, 10:25

View original

5 comments

Userlevel 7
Badge +20

Unsure, but check your logs, this could be a tenant configuration with AWS preventing connectivity, could have a firewall rule issue. What version of Veeam are you running as that’ll relate to how up to date it’s AWS integration is.

 

Logs are gonna help a lot, might find a strange IP issue here

Userlevel 2

Hi @MicoolPaul , thanks for getting back to me.

It’s not firewall or stange IP, it’s purely the AMI that’s chosen by Veeam for the proxy instance. To test, I set the termination protection on the proxy instance, so Veeam couldn’t tear it down after trying 10 times, and after a few minutes I was actually able to ssh into the instance. So it’s really just that, takes some time for the ssh service to start.

In the logs I see:

Getting x86x64 images by name: 'ubuntu-bionic-18.04-amd64-server-20180522-dotnetcore-2018.07.11';
Getting x86x64 images by name: '*ubuntu-bionic-18.04-amd64-server*';
Getting x86x64 images by name: '*ubuntu*';
Chosen AMI for AWS proxy: 'ami-00e62671e21dc4aad'

 

The chosen AMI is this one: aws-parallelcluster-3.1.4-ubuntu-2004-lts-hvm-x86_64-202205121006 2022-05-12T10-09-32.380Z, and I wonder why that is. Can that be changed somehow? I tested running the exact same AMI in the same subnet, security group etc and it really takes 5 minutes until I can ssh.

Anyway, this AMI seems wildly inappropriate for this use case?? A “normal” Ubuntu installation should do, no? No need for parallelcluster?

Userlevel 2

Ok, this seems to work better now, no idea why but I’m rarely getting timeouts now. Everything is exactly the same… Maybe was just AWS having a bad day.

 

Userlevel 7
Badge +20

Glad your problem is sorted, sucks we didn’t have anything definitive as a resolution, but as you say, likely a bad cloud day…

Userlevel 2

I still think the queries to select the ubuntu AMI should be changed, they’re not gonna return anything ever:

Getting x86x64 images by name: 'ubuntu-bionic-18.04-amd64-server-20180522-dotnetcore-2018.07.11';
Getting x86x64 images by name: '*ubuntu-bionic-18.04-amd64-server*';
Getting x86x64 images by name: '*ubuntu*';

There’s just nothing for the first 2 queries, so it will always fall back to ANY image with ubuntu in the name.

Comment