Closed linsite closed 3 years ago
We aren't doing inline virtual disk conversion yet. If your original virtual disk is 4 GiB then you have to have 8 GiB available in the storage class. The virtual disk gets uploaded into pvc then it gets optimized/converted.
Can you check your hostclass and make sure you have 8 GiB?
The PVC provided in the storage-setup.yml is 4GiB, I change it to 5GiB and the problem is solved. I guess this is not an issue, but the provided storage-setup.yml in lab needed to be updated to larger capacity, or the tutorial should be updated to check the PVC is large enough before creating the PVC for a better experience. If the advice is unnessasery, just close this issue :). I just start learning KubeVirt, so the advice may not be proper.
I believe you are right and the requested storage should be updated to accomodate the bigger image size.
https://github.com/kubevirt/kubevirt.github.io/blob/source/labs/manifests/pvc_fedora.yml#L8
Another option is to change the Fedora img because it's getting obnoxiously large. This is a reason that Alpine, Cirros, Ubuntu dominate containerized workloads.
Well, the Fedora image itself is relatively small (e.g. <200 MB vs 500+ MB for Ubuntu AFAICT from a quick look), but it as a virtual size of 4GiB that overflows the pvc by just a bit.
IMHO just increasing the requested PVC size is a simple straightforward change that will fix that and I'd vote for this (e.g. no need to update any references).
Is this a BUG REPORT or FEATURE REQUEST?:
What happened: lab2 failed
What you expected to happen: importer pod should succeed.
Anything else we need to know?: