Migrating 98TB VM

[stray_guru]

Hello all

I'm trying to migrate powered off VM containing two Virtual Disks in sum capacity 98TB. From  ESX 6.7 to ESX 6.5U2 by Vcenter 6.7. - Essential Licence. And I'm getting error "An errore  occured while comunicating with remote host"

Same while clonning. I will mention that smaller VMs migrate without any problem... Any idea?

[dbalcaraz]

vMotion won't work if you surpass the heartbeat timeout between ESXi hosts so, maybe (I don't how did you configure your storage) your network has some latency from a particular ESXi host and hence, denying the vMotion.

Did you check in the vxpd.log in order to see a detailed error of this?

[stray_guru]

this is my vpxd log I just cut only part of it, since log-timestamp "start to completion time" of failed job. Can You help me debuging it?

[nestorleal641]

Check that you do not have a snapshot, you can perform the migration with some software such as Veeam Backups & Replication

[continuum · Accepted solution]

If the task is: copy 100TB from ESXi 1 to ESXi 2 and you expect problems along the way you first of all need a no-nonsense approach that can be restarted when ever it fails.
Having to restart from the beginning after 90% is done is no allowed option when the whole process can take days and  days.
This is what I do ....
start a Linux VM on one of both hosts
boot that VM and connect to source datastore on ESXi 1 and target datastore on ESXi 2 via sshfs.
mount both datastores like this
sshfs -o ro root@source-esxi:/vmfs/volumes/datastore-source/ /vmfs-in

sshfs root@source-esxi/vmfs/volumes/datastore-target/vmfs-out

then in a first batch copy all small files other than the large flat.vmdks with a command like
cp /vmfs-in/directory/* /vmfs-out/directory/
this will just take a few minutes.
To copy the large flat.vmdks I use ddescue - as that offers the option to restart when ever necessary
ddrescue /vmfs-in/directory/name_1-flat.vmdk /vmfs-out/directory/name_1.flat.vmdk  /vmfs-out/directory/name_1.flat.log

ddrescue /vmfs-in/directory/name_2-flat.vmdk /vmfs-out/directory/name_2.flat.vmdk  /vmfs-out/directory/name_2.flat.log
once both commands are started you can relaunch them when ever the commands gets interrupted - and you can count on having that option!!!
and Yes:
this is ugly
this may not be the fastest option
this may not be the option you learned in your VCP-course
but if your boss asks you "will you manage this until next friday ???"
this is the only way to do it and sleep well during next days 😉
I use this approach in my VMware recovery work whenever I have to copy very large vmdks and need predictable results.
It is ugly, requires extra work but it is predictable and reliable and is the only way that can handle any network problems along the way.
If source or taget ESXi reboots because a powerfailure tomorrow this approach can continue the copy at the place where it was interrupted.
This approach surely is overkill for small VMs - but in your case investing an extra 30 minutes to set this up is a very reasonable decision.
Just make sure that you use the "create log" option with ddrescue !!!

[stray_guru]

hello

yes this is perfect solution I was thinking of. But before this final solution i checked Veeam migration option which perfectly bypass the problem. Maybe not perfectly because much slower than ekspected ;( but til now it's copying. However I'm still curious why it won't work in vcenter-vmotion

[stray_guru]

no snapshots... yes I switched to veeam but it's slower solution and I'm still curious why it won't work

[dbalcaraz]

I was thinking... are you trying to move only from a compute perspective or also the storage (different Datastores) ?

About the vxpd.log, I only found this error: -- ERROR lro-29078 --  -- VmprovWorkflow: vmodl.fault.HostCommunication:

[stray_guru]

I'm moving also storage so simply it's copying 98TB of data

[dbalcaraz]

What  continuum said is nice.

Another thing you could try is to SDRS disks, instead of doing the whole VM.

[jlaurentino]

The solutions provided seem good.

I have only 7 TB in my ESXI setup and have an offline redundant machine.

Every so often I do the backup by going on a workstation and

mkdir esxi-from

mkdir esxi-to

sshfs main-machine esxi-from

sshfs backup-machine esxi-to

sudo rsync -aAXHv esxi-from/ esxi-to/

Any suggestions to improve this would be greatly appreciated.

Thanks,

Laurentino