error backing up VMs from vSphere 5.5 (please don't shoot at me)

[tinto1970]

hi folks,

I've been asked to migrate VMs from a vSphere 5.5 infrastructure, not recieving updates since 2014 (!) to a 7.0.

Theoretically Veeam 11 should be compatible with 5.5 but the backups wasn't running. Error is, briefly "Failed to open disk for read.".
I had in my archive an old Veeam 9.5 iso and I tried to install it (community edition could be enough for me). Sadly I recieve the same error, shown below in detail.
Yes, I forced NBD mode.

I'm pretty sad, has someone an idea about the reason for this error? I also have a Veeam 8 iso in my archive, maybe I could try this too?
Thanks in advance

Error: Failed to open VDDK disk [[DX200_DB01] REDHAT65/REDHAT65.vmdk] ( is read-only mode - [true] )
Logon attempt with parameters [VC/ESX: [10.31.12.150];Port: 443;Login: [root];VMX Spec: [moref=vm-87];Snapshot mor: [snapshot-1232];Transports: [nbd];Read Only: [true]] failed because of the following errors:
Failed to open disk for read.
Failed to upload disk.
Agent failed to process method {DataTransfer.SyncDisk}.

[MarkBoothmaa]

How about using the Agent and do instant recovery and then migrate to production?

[tinto1970]

I had this idea, or to use VMware converter, but the operating systems (RHEL 6) are as old as the infrastructure, and it's really risky to install anything on them. There is a total lack of maintenance, maybe incredible uptimes, lack of disk space etc etc.

[david.domask]

Hi tinto1970,

Regrettably we won't be able to troubleshoot it over the forums and given the out of support nature of the environment, agents were going to be my suggestion as well. I'm not sure older Veeam versions will be the answer here personally as seems like a VDDK error, so it would require log investigation to solve.

Similarly, I imagine VMware's VMware converter in v2v mode could do this with downtime and a bit of scripting, but I get your caution on this.

If you want to try to investigate the logs yourself, check the source agent vmdk logs for the job that failed, usually the errors there are fairly clear as it's about the datamover agent talking to the VMware environment through VDDK library. For issues like file locks, connectivity errors, permissions, etc, the errors are usually quite clear.

[MarkBoothmaa]

File locks is a good shout especially if the hosts have been up for a long time. You could check to see of one of the VM's will vMotion to another host.

[tyler.jurgens]

Solid callout there Mark. I'd look to do the same thing - vMotion elsewhere. Or try to shut down the VM and move it to another host while the VM is in a powered off state. If you can't get that VM or others off the host, you can try to shut down the VM (and all VMs on the host) and reboot. I'd look to isolate if its just a single VM or multiple VMs on that host as well.

[tinto1970]

thanks everyone, folks. your help is unvaluable, nice to be a community

no, i didn't solve yet, but I'm not feeling alone, and it's great

[mjr.epicfail]

I would install 6.x to one migration host, move everything of 5.5 tot 6.0, then got to 6.7 and to 7.0
Is a pain but to my knowledge the only route to keep everything online and being able to vmotion (not sure if vmotion would work between 5.5 and 7 or 5.5 to 6.7

[tinto1970]

thank you, good point, I forgot to mention that I'm not allowed to install/move anything on the previous infrastructures, and to spend such an effort on the project

[ThorstenS]

As a linux guy, I would boot a clonezilla.iso, save the VM and restore them to a freshly created VM in the new infrastructure.

Or - if you have the balls:
shutdown every daemon on an old VM and use pv or dd (with bs=2k), pipe it to netcat with netcat server listening on a freshly created VM, running a live system, and write it with dd to the disk.
I did this ~15 years ago, when migrating several VMs from XEN to KVM. Force a fsck on first boot.
And don`t forget to skip uefi on the new machines

hth

EDIT:
Here is an old static netcat binary - If you don`t have it on the machine:
https://github.com/andrew-d/static-bina ... 86_64/ncat

[mjr.epicfail]

If your not allowed to spend the time to be carefull and thorough, then I hope those are replaceable servers

[EdFromOhio]

thank you, good point, I forgot to mention that I'm not allowed to install/move anything on the previous infrastructures, and to spend such an effort on the project

With those restrictions, it sounds like your decision makers actually want these old systems to fail. If it were me, I'd also be in the vMotion camp, but would be looking at trying to get to the latest version instead of 7.0. If you're going to update it, why not all the way since they seem to not touch it for a decade at a time?

[roland]

Do you try this ? https://community.broadcom.com/vmware-c ... -vddk-disk
'installed the Microsoft Visual C ++ 2008 Redistributable Package (x64), then I performed a backup and it worked.'

[Matts N]

With those restrictions, it sounds like your decision makers actually want these old systems to fail. If it were me, I'd also be in the vMotion camp, but would be looking at trying to get to the latest version instead of 7.0. If you're going to update it, why not all the way since they seem to not touch it for a decade at a time?

A very valid point. Probably management doesn't care about having a supported environment, but if they do... vSphere 7 is end-of-general-support in October this year.

[tinto1970]

Do you try this ? https://community.broadcom.com/vmware-c ... -vddk-disk
'installed the Microsoft Visual C ++ 2008 Redistributable Package (x64), then I performed a backup and it worked.'

thanks! I tried both on Veeam 11 and 9.5 installations, but the backup still fail with same error message

[tinto1970]

As a linux guy, I would boot a clonezilla.iso, save the VM and restore them to a freshly created VM in the new infrastructure.

Or - if you have the balls:

I love the linux guys, and their balls
but the OVF export / import appears to be the easier way at the moment

[tinto1970]

an angel suggested me to try installing a Ubuntu proxy on the old 5.5 infra, and see what happens with HotAdd transport.

I'll give it a try in case I get the green light to spend more time resources on it.

P.S.: the "political" situation is pretty complicated, noone wants systems to fail, but there is a pretty long "supply chain", we don't deal directly with the customer, and the customer is a complicated organization. Where something is actually working not as good as it should: the total lack of care of the systems is a proof of that.

[msee]

Actually I had same situation last year with ESXi 5.5 and VM migration to ESXi 8.
Solution was to install vCenter 6.0 - used Windows server based one not appliance, to connect ESX 5.5 hosts to it and then to connect latest Veeam Backup to this vCenter 6.0.
Veeam Proxies were normal Windows 2022 servers that I did add on these ESXi 5.5 machines and whole setup worked very well - got 60+ VMs and several TB of data migrated in no time.