VMware ESXi import problem
-
I've been slowly migrating VMs from ESXi and most of the time the V2V feature works very well, I was able to migrate about 30 vms without any problem except I have 2 cases where it didn't work so let me explain one of them.
I have a Windows 2022 server VM installed on a ESXi v7.x server that I wanted to import via V2V. To facilitate things, I stopped the VM in ESXi. The image size is about 350gb. When I start the V2V import I see the jobs being created in XO but after 1 hour it is still @ 0% and shows 27 days for import time. I killed the job.
I decided to import the VMDK disk itself and later attach it to a already created VM. I downloaded the VMDK file from the ESXi server via winscp with no problem. I used the Import/Disk feature of XO but as soon as I select it (or drop it in the window) in the import tool, I see the 'turning' logo appearing bollow the drop window and it stays like that forever. I reloaded that page after 4 hours and nothing was done.
This one is the biggest vm in terms of size that I need to import so I'm not sure it it's related to size or not. Since I'm not getting any error message I can't say that they fail but both of the import never ends.
XO VM image from source on github running on XCP-ng 8.21 with latest patches. HW is Dell R730XD. I was able to import other VMs via the same XO and server.
Any ideas as to what I should try next?
Thanks.
JF.
******** Update 2024-09-04 ********
I was finally able to migrate that VM. The solution that worked for me was not in the docs nor in the suggestions in the replies as I was not able to apply them. Since I was able to import other VMs but not this one and importing the VMDK disk didn't work either I decided to see if it had something to do with the VMDK itself. I dusted off a tool that I used for migrations/backups a long time ago... The ol' VMware Standalone Converter to convert from inside the running VM. This process is unaware of any snapshots or disk consolidation issues (could have be the culprit) and will simply generate a completely new and shiny VM from the running one.
I used another ESXi server as the destination but I could have used the same one as the source. Immediately after the Standalone Converter was done, I was able to start the V2V import process from XO to import the newly generated VM and it completed in about 40 minutes for a 350gb VM. I had to tweak the network a little like it is almost always the case and voilà!
Now I will play a little with only other import failure (out of 30 ish successful ones) and make another post about it if I don't succeed because it seems to be a very different problem than this one. I only leave the solved comment here in case it may help someone in the future
-
As all my other attempts have failed so far, I'm trying the direct conversion. Following the procedure, I'm able to transfers the VMDK in the XCP-ng server using WinSCP but when I try to install the packages needed to do the conversion, I get an error message on the first command from these:
yum install qemu-img --enablerepo=base,updates wget https://forensics.cert.org/centos/cert/7/x86_64/vmfs6-tools-0.2.1-1.el7.x86_64.rpm yum localinstall vmfs6-tools-0.2.1-1.el7.x86_64.rpm
The error message is:
Could not retrieve mirrorlist http://mirrorlist.centos.org/?release=7&arch=x86_64&repo=os&infra=$infra error was 14: curl#6 - "Could not resolve host: mirrorlist.centos.org"
From that machine, using SSH I can ping the whole internet including 'centos.org' but 'mirrorlist.centos.org' doesn't no resolve. It doesn't resolve on other computers as well. I lack the necessary knowledge/courage to be 100% sure what to do next and I especially don't want to mess up the server! I also tried that command on another XCP-ng server in another location with the same result.
Any idea what I could be doing wrong ?
Thanks.
JF.
-
Check the logs.
journalctl -u xo-server -f -n 50
-
@jflaplante you could try creating an ova, that worked in some cases for us.
-
The other option is a direct vmdk convert to VHD, see https://docs.xcp-ng.org/installation/migrate-to-xcp-ng/#convert-a-vmdk-file-to-a-vhd
All in all, the entire doc provides many solutions: https://docs.xcp-ng.org/installation/migrate-to-xcp-ng/#-from-vmware
-
@jflaplante Thanks for the suggestions, I will be trying them when I work on that server again.
I guess I was more fixated by the fact the I was unable to upload any data in both of my attempts as It felt like the door was closed and gave no error message.
I will update one way or the other.
Thanks again.
JF.
-
As all my other attempts have failed so far, I'm trying the direct conversion. Following the procedure, I'm able to transfers the VMDK in the XCP-ng server using WinSCP but when I try to install the packages needed to do the conversion, I get an error message on the first command from these:
yum install qemu-img --enablerepo=base,updates wget https://forensics.cert.org/centos/cert/7/x86_64/vmfs6-tools-0.2.1-1.el7.x86_64.rpm yum localinstall vmfs6-tools-0.2.1-1.el7.x86_64.rpm
The error message is:
Could not retrieve mirrorlist http://mirrorlist.centos.org/?release=7&arch=x86_64&repo=os&infra=$infra error was 14: curl#6 - "Could not resolve host: mirrorlist.centos.org"
From that machine, using SSH I can ping the whole internet including 'centos.org' but 'mirrorlist.centos.org' doesn't no resolve. It doesn't resolve on other computers as well. I lack the necessary knowledge/courage to be 100% sure what to do next and I especially don't want to mess up the server! I also tried that command on another XCP-ng server in another location with the same result.
Any idea what I could be doing wrong ?
Thanks.
JF.
-
Old CentOS repo changed, we have an update coming soon to solve that. @stormi when do we have that? Next week?
-
-
-
This specific update is ready, but the next update train (we group updates together to avoid unnecessary maintenance) is not scheduled yet.
-
The update in question can be cherry-picked without notable risk from the testing repository:
yum update xcp-ng-release --enablerepo=xcp-ng-testing,xcp-ng-candidates