https://rancher.com/ logo
Join Slack
Powered by
This message was deleted.
# harvester
a
This message was deleted.
g
this is a known issue and being tracked: https://github.com/harvester/harvester/issues/3095
we have the fix ready.. and we are trying to get it in v1.1.1
q
@great-bear-19718 that's great news! i saw the other error about how it doesnt import if the vm doesnt exist in the new cluster. didnt see this issue. thanks! does this fix address the backing image as well?
if not, any suggestions on exporting backing image from 1.0.3 > v1.1.0? (or v1.1.1 for that matter)
g
it will only allow existing vm backups to be imported into a new cluster
so are you trying to export backing image and then import it in a new cluster?
q
kind of. i have a v1.0.3 cluster. a lot of my vm's are ports from hyper-v, so there's a backing image from when i exported from hyper-v, converted to qcow, then downloaded into the cluster.
i've noticed that if i take a backup of one of those vms (backing image + changes since i got it up and running in harvester) that the backup is super-small. so i'm assuming the backing image is not part of the vm backup.
so if i backup / restore, i'm still going to run into an issue of getting the backing image from old cluster -> new cluster. https://github.com/harvester/harvester/issues/3016
here's how i migrated from hyper-v -> harvester if that helps explain my process: https://github.com/harvester/harvester/issues/2950
g
the import can only happen from a http endpoint so when you export the image will need to be made accessible via a http endpoint
we introduced a vm-import-controller in v1.1.0, we do want to add harvester as a source so we can do the migrations from one cluster to another
q
yeah, i get that, now. but i have a whole slew of images i downloaded into harvester from an http when i was migrating. but as i imported them, i removed them from the server (i had to migrate like 30 vms) so now i have no way of getting them from old cluster to new cluster. does that make sense?
g
but i dont have an ETA for it, likely 1.1.2
yeah... makes sense
the best option would be to wait for 1.1.1 and use the nfs backup to repopulate a new cluster
q
got ya. but am i right that the backup does not backup the backing image as well?
and if that's the case, any thoughts on how to get that out of cluster A, and into cluster B?
g
let me check
q
p.s. you people are amazing! 😄 i love this project.
g
i dont have a 1.0.3 cluster running but are you able to access longhorn and see the backing images there?
q
yes
g
then the only option right now may be to download the backing image from longhorn and upload it using the file upload in harvester
q
i'd be fine w/ that. how do i find it to download?
*in longhorn
g
the image name would be the same as the virtualmachineimage name in harvester
if you view yaml of an image you want to export you will see the name in 
metadata.name
it is randomly generated and different from the display name you see in the UI
that will map to $namespace-$imageName in longhorn backing image
q
okay, so here's an example:
it show's i puled it form that ip, it's obviously no longer available there like we discussed...
so how do i get that out of longhorn?
g
that is fine
on the right you should see download option in the menu item
q
no 😞
just delete and cleanup
(under operation)
g
aah.. thats what i was trying to check first 😄
it must be pinned to version of LH
q
yeah. my guess 😞
g
i will check LH release notes
q
i really appreciate it!
this is 1.2.4 btw (lh)
g
👍
that bit i know from harvester release notes 🤣
q
at least you got me in a direction, i appreciate it.
lol
g
q
so i guess i'll have to rifle through lh docs and see if i can figure out how to get an image out of it on older versions.
g
i will ask internally
q
i'd almost rather manually move vms from old cluster to new cluster for now, instead of brave an upgrade once they fix the issues... i'm using harvester in prod, so being down for a day or two while working out upgrade bombs would be an issue.
thanks! i appreciate it!
(again)
g
1.1.1 should be out in a couple of weeks
q
careful. you might become my new favorite friend! 🤣
g
so unless you are waiting for some feature only available in 1.1.x i'd wait
q
i'm at least waiting for that. for sure.
g
then you should be able to upgrade in place and also migrate using backups
q
and yeah, i want storage classes in 1.1 so bad it hurts!
62 Views
Open in Slack
PreviousNext
Powered by