Currently I have a One Cluster made of a bunch of compute and at least 3 kinds of datastores :
DS with ssd
DS with fast drives
DS with slow drives
I want to achieve the following : I want to make a template VM which will be deployed automagically with the system partition on DS with fast Drive, data partition (files, archives, …) on slow drives
For the moment, I can’t see how to deploy an image to a specific datastore and as well, how to add a disk to an already instanciate VM on a specific datastore.
I am stucked on this for some times now, I read and re-read a bunch of documentation but I don’t know what I am missing
Thanks for your answer.
Unfortunatly, I already knew that and it doesn’t help me.
As far as I see, I can’t deploy a VM with a system on a fast DS and add a disk on a slow DS.
If I understand well, I can’t either set an affinity to a DS for an image.
What do you mean “add a disk”? A generic storage datablock image or a volatile disk? Because you can certainly add a datablock image instance and attach it to a VM, from whatever datastore the image is created under.
Hi. Actually I accidentally got what you trying to get
I have several LVM datastores, and when I moved VM from one DS to another all volatile disks stayed on old one.
So now I’m having OS disk on one DS and Data disk on other.
It’s looks like a bug in tm_mad.
But you can manually move desired LV to another DS with dd and make symbolic link. This works either.
Looking at the TM_MADs it looks like the same use case as How to use a second datastore but for fs_lvm driver - the images are transferred to the SYSTEM datastore but you need to have symlinks to their datastores instead… (note that you haven’t the luck to use “persistent” images as workaround, though.)
Lets hope it will be rainy outside so will have some more spare time…