VMware migration tool: we need your feedback!

olivierlambert

Are you sure it's not a PIF? Can you do a xe network-param-list uuid=<UUID>?

If it doesn't work, then do it for a PIF: xe pif-param-list uuid=<UUID>

brezlord

@olivierlambert said in VMware migration tool: we need your feedback!:

xe network-param-list uuid=

yes you are right I had an error in the uuid. I copied it from the host and not the pool.

olivierlambert

Does it work now? In XO6, we'll do everything to avoid the confusion between a PIF and a network, it should be more clear than it is today.

brezlord

@olivierlambert yes. I very excited to test out XO6.

planedrop

Just as some additional feedback about this, I tried with thin=true today on a small 20GB Ubuntu VM and it worked great!!

I do have a suggestion though, I'd love to see a task in XOA about reading the blocks from the ESXi VM. When I entered the command, I thought nothing was working because a task never started, but when I checked network stats on the host it was very clear it was reading from ESXi. Once it finished reading, it imported the disk (which created a task) and the actual VHD space used is only 7.5GB!

olivierlambert

Yes, as I explain, the first pass is just reading the whole VMware disk once. It's only doing the transfer on second pass.

That's why there's no XCP-ng task in the first pass (since until now, we only have XCP-ng tasks, there's no XO tasks).

But it's changing, we are about to release XO tasks, so it will be easier to track the job

planedrop

@olivierlambert Totally makes sense, this is great news.

Thanks again! I'll be testing the tool on some very large VMs in the coming weeks so I'll report back if anything weird happens with those.

brezlord

@florent When I import a windows 10 VM from ESXi to XCP-ng the boot firmware was incorrectly set to bios when the source VM was uefi. I would also be advantageous to have the mac address(s) copied over by default or at least on option.

Thanks,
Simon

florent

@brezlord nice catch, I will work on it today

@planedrop : task progression is in the work with @julien-f , and I hope they'll reach maturity soon

florent

@brezlord mac address and uefi should works now

brezlord

@florent I will rebuild XO and re-import and give feedback. Thanks.

olivierlambert

It's now available directly on master (from the sources) or on latest XOA release channel (I updated the first post accordingly)

planedrop

@olivierlambert Awesome, super exciting stuff!

brezlord

@florent Everything is working now, Thanks. I have sucsefully migrated Windows and Linux VMs.

olivierlambert

Perfect! Now expect all of this but in a simple wizard in the UI, that will be a great tool for everyone who want to make the migration (and we hope to get a decent share of those users with our new support bundle)

severhart

This post is deleted!

severhart

After running for hours, get the following

✖ sesparse Vmdk reading is not functionnal yet FP-FileServer/FP-FileServer-000001-sesparse.vmdk
JsonRpcError: sesparse Vmdk reading is not functionnal yet FP-FileServer/FP-FileServer-000001-sesparse.vmdk
    at Peer._callee$ (/usr/local/lib/node_modules/xo-cli/node_modules/json-rpc-peer/dist/index.js:139:44)
    at tryCatch (/usr/local/lib/node_modules/xo-cli/node_modules/@babel/runtime/helpers/regeneratorRuntime.js:44:17)
    at Generator.<anonymous> (/usr/local/lib/node_modules/xo-cli/node_modules/@babel/runtime/helpers/regeneratorRuntime.js:125:22)
    at Generator.next (/usr/local/lib/node_modules/xo-cli/node_modules/@babel/runtime/helpers/regeneratorRuntime.js:69:21)
    at asyncGeneratorStep (/usr/local/lib/node_modules/xo-cli/node_modules/@babel/runtime/helpers/asyncToGenerator.js:3:24)
    at _next (/usr/local/lib/node_modules/xo-cli/node_modules/@babel/runtime/helpers/asyncToGenerator.js:22:9)
    at /usr/local/lib/node_modules/xo-cli/node_modules/@babel/runtime/helpers/asyncToGenerator.js:27:7
    at new Promise (<anonymous>)
    at Peer.<anonymous> (/usr/local/lib/node_modules/xo-cli/node_modules/@babel/runtime/helpers/asyncToGenerator.js:19:12)
    at Peer.exec (/usr/local/lib/node_modules/xo-cli/node_modules/json-rpc-peer/dist/index.js:182:20)

florent

@severhart the sesparse format is used for disk greater than 2TB , and for all disks after esxi 6.5 , and it's not very documented
https://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.vsphere.storage.doc/GUID-88E5A594-DEBC-4662-812F-EA421591C70F.html
we are working on implementing this reader to allow migration during Q1 2023

For now esxi 6.5+ vm are limited to migration of sopped vm without snasphots

severhart

@florent thanks! for the fast reply, I will take the outage and rerun, and let you know outcome.

olivierlambert

Also, 2TiB+ disk can't be imported since it's limited to 2TiB tops on XCP-ng default storage stack.