Dear All,
I have been using ESXi 3.5 (free license) for sometime and utilise the 'linked clone' script available here: http://communities.vmware.com/docs/DOC-9202 I understand from many posts on these forums that VMware have now patched/fixed the ability for RCLI write access and so the aforementioned script no longer functions. Whilst this is an understandable move from VMware to protect its products, it does cause me some issues as I really value the ability to create a linked clone and therefore very quickly deploy a new VM.
From my own research I have a workaround, (its not great but works), basically the following script available here http://engineering.ucsb.edu/~duonglt/vmware/ (note not my work) performs a linked clone in a few steps (note brief and simplified overview):
1) Copy the master VM's .vmx and .vmxf configuration files into the new location.
2) Edit the new .vmx config file to point back to the master VM's virtual hard drive.
3) Register the new VM (basically adding it to the ESXi inventory).
4) Create a snapshot (using the new .vmx config file), therefore creating a new virtual hard disk to store the changes (or differences).
I personally think the author of the script deserves a lot of credit as the technique is very neat.
In ESXi 3.5 U4 the features within the RCLI used to perform steps 3 and 4 are no longer available as they require write access. My solution is to run steps 1 and 2 from the script (as before) followed by using the VIX API to register the VM and create a snapshot. This works, however does require running two seperate programs. When I first looked into using the VIX API I noticed a function called 'VixVM_Clone' which is capable of creating full clones and linked clones, but guess what - yup thats right this is not supported in ESX or ESXi. Note the VIX API support matrix here: http://www.vmware.com/support/developer/vix-api/vix16_reference/supportMatrix.html
I hope the above workround will help others and please if someone has a better solution, perhaps they would be kind enough to post it......
I have been using ESXi 3.5 (free license) for sometime and utilise the 'linked clone' script available here: http://communities.vmware.com/docs/DOC-9202 I understand from many posts on these forums that VMware have now patched/fixed the ability for RCLI write access and so the aforementioned script no longer functions. Whilst this is an understandable move from VMware to protect its products, it does cause me some issues as I really value the ability to create a linked clone and therefore very quickly deploy a new VM.
From my own research I have a workaround, (its not great but works), basically the following script available here http://engineering.ucsb.edu/~duonglt/vmware/ (note not my work) performs a linked clone in a few steps (note brief and simplified overview):
1) Copy the master VM's .vmx and .vmxf configuration files into the new location.
2) Edit the new .vmx config file to point back to the master VM's virtual hard drive.
3) Register the new VM (basically adding it to the ESXi inventory).
4) Create a snapshot (using the new .vmx config file), therefore creating a new virtual hard disk to store the changes (or differences).
I personally think the author of the script deserves a lot of credit as the technique is very neat.
In ESXi 3.5 U4 the features within the RCLI used to perform steps 3 and 4 are no longer available as they require write access. My solution is to run steps 1 and 2 from the script (as before) followed by using the VIX API to register the VM and create a snapshot. This works, however does require running two seperate programs. When I first looked into using the VIX API I noticed a function called 'VixVM_Clone' which is capable of creating full clones and linked clones, but guess what - yup thats right this is not supported in ESX or ESXi. Note the VIX API support matrix here: http://www.vmware.com/support/developer/vix-api/vix16_reference/supportMatrix.html
I hope the above workround will help others and please if someone has a better solution, perhaps they would be kind enough to post it......
