Veeam 6 and ESXi 5 - SSH Failed To Connect

[bunger]

I am running Veeam 6 and just upgraded my 3 ESX hosts to ESXi 5. I am noticing that my backup jobs are all failing with SSH connection errors. Here is the error:

Failed to login to "myEsxiHost1" by SSH, port 22, user "root", elevateToRoot False, autoSudo False
User authentication failed

I am able to connect to the ESXi host with the same credentials via a SSH client ( putty ), so I know the credentials are good and that SSH is running.

Any thoughts or suggestions would be very much appreciated!

tia,
Bill

[bunger]

A quick follow-up: the issue seems to be with the Backup Repositories. My backup jobs are configured such that they backup to each of the others' ESXi datastore and the Backup Repository wizard is failing the connect to the ESXi host....

[Vitaliy S.]

Hi Bill,

My backup jobs are configured such that they backup to each of the others' ESXi datastore and the Backup Repository wizard is failing the connect to the ESXi host....

ESXi host cannot be used as destination targets for the backup jobs as per our system requirements.

See this topic for more details: ESX(i) host not listed in destination for backup

As to your error message, then looks like Veeam B&R is trying to access your ESXi host as Linux server, which was possible with "fat" ESX, but no longer an option with ESXi 5.

Thanks!

[bunger]

The VMs are all stored on a SAN and my backup jobs took those VMs and created backups on each of my 3 ESXi 5 hosts. The hosts were running ESX 4 before the upgrade to ESXi 5. I have over 2TB of storage on each ESXi host for this purpose.

What are my "new" options?

Thanks,
Bill

[Gostev]

One of the supported backup targets, as per our System Requirements (see the Release Notes document included with your download). For example, VM with disks stored on these local datastores. Although I would not recommend backing up to VMFS, as this has always been considered a bad practice (search the forum for existing discussions on this). I would definitely recommend getting proper backup target instead, there are a few good topics around talking about different backup targets, for example this one is quite active right now. Thanks!

[bunger]

So just to verify:

Veeam will backup to ESX without any issues.
Veeam will NOT backup to ESXi, though that is the only option that VMWare now offers.

Am I understanding that?

Bill

[Vitaliy S.]

Yes, you're reading this correct.

[Gostev]

Veeam will backup to "fat" ESX only because those can be treated like regular Linux servers, while we support Linux servers as targets natively. ESXi is entirely different architecture though, and it does not support any 3rd party code running in its "service console".

[bunger]

Hmmmm.....

Will Veeam support replications to ESXi 5 datastores and if so, can I also restore files from a replica? Maybe that will be temporary workaround....

[Vitaliy S.]

Replication to ESXi hosts is fully supported as well as FLR operations from replicated VMs.

[Gostev]

Just be sure to have solid and tested plan on how to restore anything if your ESXi host with those replicas is fried.

Also, always keep in mind that backup is not a backup until you have 3 copies of it. For example, if the VMFS datastore holding replicas gets corrupted or lost as a part of the disaster, you will have no way to restore any data.

[bunger]

I just encountered a related issue to Veeam's inability to connect to ESXi 5:

I have a ton of backups and replicas that exist on each ESXi 5 host ( created when each host was running ESX 4 ). When I try to delete them from Veeam, I get the same connection error.

Code: Select all

Error: No connection could be made because the target machine actively refused it

How can I remove these obsolete backups and replicates from Veeam?

Thanks
Bill

[Gostev]

I suppose our support should be able to provide you with the SQL script to edit Veeam B&R configuration database directly.

[cheib]

Hi together,

i am facing the same error since i updated my esxi 4.1 environment to esxi 5 . My Veeam Backup VM is running on my Vsphere 5 Cluster and i presented several
raw device luns to this vm for backuing up my environment. now since i upgraded my hosts to esxi 5 i am not able to backup anything with exactly this error.
is there a possibility to help me out in this case ?

thanks in advance
Hardy

[Vitaliy S.]

Hi Hardy,

SSH is no longer available in vSphere 5, at least you should enable it manually, which is unsupported by VMware. Please let our support team have a look at your debug logs and apply corresponding SQL script to change the connection mode if necessary.

Thanks!

[cheib]

Hi Vitaly,

yes but this was already the case in esxi 4.1 so no new situation in esxi 5. i already enabled the shell and ssh and from putty i am able
to connect to the esxi hosts. the strange thing is that with one esxi5 host the connection is successfull but with others not ...
should i open a case with support for uploading the logs ?

[Vitaliy S.]

Yes, it would be better to take at your issue live via remote session or deeper debug logs investigation.

[cheib]

hmm still waiting for support response .....

[cheib]

OK finally i got response from support, thanks for that first.

It seems that my issue has to do with duplicate hosts entries in the veeam database.
it's simple to open sql mgmt studio and edit the dbo.hosts table and clean up the old
esxi 4.1 hosts against the new esxi 5 hosts.

[schedlbauerm]

[merged]

I am running Veeam 6 and just upgraded 4 ESX 4.1U1 hosts to ESXi 5 U1. I am noticing that most of my backup jobs are failing now.

All vms on esx1 and esx2 get backed up. All backups of vms on esx3 and esx4 don't work.

If I wan't to connect to esx3 and esx4 in the veeambackupconsole I get the following error message:

Failed to login to "myEsxiHost1" by SSH, port 22, user "root", elevateToRoot False, autoSudo False
User authentication failed

I am able to connect to the ESXi host with the same credentials via a SSH client ( putty ), so I know the credentials are good and that SSH is running.

I think there must be a mismatch of the esxversion of esx3 and esx4.

We named the new esx5iu1-hosts exactly as the previous esx4.1u1-hosts.

As described in the following forum-entry:

http://forums.veeam.com/viewtopic.php?f ... 43&start=0

I think my issue has to do with duplicate hosts entries in the veeam database.

Perhaps I should open sql mgmt studio and edit the dbo.hosts table and clean up the old
esxi 4.1 hosts against the new esxi 5 hosts.

But I need clarification if this might solve the problem and how exactly I have to edit the dbo.hosts table.

Best regards,

Martin

[Gostev]

Hi Martin, for this kind of assistance you need to open a support case please, and one of our support engineers will help.

[rbagby]

I just wanted to add that I was having the same exact problem. Going from ESX 4.1.x to ESXi 5.0 U1 with a fresh install and reusing the same name and IP. I received the SSH Failed to connect in Veeam but I could SSH to each host without issue. Schedlbauerm was correct in that the Veeam database table dbo.Hosts listed both the new host on ESXi 5.x and the old host on ESX 4.x.

I recommend working with support to resolve. If you decided to modify this yourself I highly recommend that you backup the Veeam SQL database first. Once the changes are made you can restart the Veeam Backup Service and connect to each host without incident.

Thanks,
Randy

[pizzim13]

I have seen this same issue with Veeam 6.1x not removing old host entries from the host table in the database. Not only did I have dupe entries for ESX/ESXi hosts it had an old entry for a repo. The "deleted" host entries are not visible in the gui but can be retrieved with powershell.

[Gav@GH]

[merged]

Testing at my site with 5.1, but stumped at the first issue of connecting with host - as there is no longer a Service Console the VMware server propoerties don't have any connection options.

It does not go through a wizard and just skips straight to a summary (Username has been specified at vcenter level):

VMware ESXi Server 'esx51.mydomain.com.au' was successfully modified.
Host info: VMware ESXi 5.1.0 build-799733.
Connection options:
User:
Port: 443

Jobs are just failing (ESX firewall is open for ssh22):

24/09/2012 5:23:46 PM :: Error: No connection could be made because the target machine actively refused it 192.168.xx.xx:22

Anyone got past this first hurdle?

[Gav@GH]

Just to clarify, I'm trying to connect to the target ESX host.

Also, no issues connecting with the same credentials using the built in putty tool.

[Gostev]

Gav@GH, your issue is not specific to vSphere 5.1... "fat" ESX was dropped by VMware back in vSphere 5.0 release, which is when users started to run into this problem during host upgrades involving host type change ("fat" ESX to ESXi) - see above.

Basically, when you are upgrading "fat" ESX to ESXi, some Veeam database cleanup is required to make Veeam aware of host type change (to prevent it from attempting to SSH to ESXi host). Before you do this, Veeam will not know that the host type has changed, and will continue trying to connect via SSH. Our support can assist you with cleaning this up.

Please, read this topic for more details and similar reports.

[Gav@GH]

Thanks for the clarification and the merge. The only reason I posted here was because support would not offer any assistance as soon as they saw I was talking about 5.1. I'll try an follow up again.