Importing qcow2 question I have a 150GB qcow2 image converte rancher-users #harvester

Title

proud-addition-21854

11/04/2022, 7:27 PM

Importing qcow2 question. I have a 150GB qcow2 image converted from vmdk . When I import it using file i get “Context Canceled” error after 99% of uploading. Using URL i see the error “Pod spec node ID host3 doesn’t match the desired node ID host3" after 99%. How do I debug what is wrong?

great-bear-19718

11/06/2022, 11:35 PM

are you please able to generate a support bundle when this error is generated

the image would likely stay in error state..

proud-addition-21854

11/07/2022, 5:31 PM

Yes I just generated it. Which log file should I look?

Ok I see the below error in longhorn-system/longhorn-manager-648xd

level=error msg="Backing Image Data Source was state in-progress but the pod became not ready, the state will be updated to failed, message: pod spec node ID host1 doesn't match the desired node ID host3 backingImageDataSource=default-image-zxgnk controller=longhorn-backing-image-data-source diskUUID=588588f4-1bec-41c8-9fde-936d5a7bf492 node=host3 nodeID=host3 parameters="map[url:<http://10.1.1.43/vm1.raw>]" sourceType=download

great-bear-19718

11/07/2022, 10:24 PM

are you able to please DM me the support bundle?

quaint-alarm-7893

11/08/2022, 5:37 PM

Paras, i ran into that too, basically you have to download, instead of upload. i used a simple nginx server, worked great! i wrote a few writeups about migrating from hyper-v to harvester, i used similar techniques for moving vms from KVM -> harvester as well. hopefully they are helpful to you. https://github.com/harvester/harvester/issues/2950 https://github.com/harvester/harvester/issues/2949

proud-addition-21854

11/08/2022, 9:13 PM

@quaint-alarm-7893 I did use the download using a web server locally. Thats how I got the “does not match” error

quaint-alarm-7893

11/08/2022, 9:16 PM

what web-server? i know i had issues with apache and iis, nginx was super-easy and ultimately i used that.

proud-addition-21854

11/08/2022, 9:16 PM

apache

quaint-alarm-7893

11/08/2022, 9:18 PM

maybe try give nginx a try? idk if it's an option for you. i know iis and apache were a pain w/ large files for me.

for me i basically turned off apache, installed nginx from package-manager (for me it was ubuntu, so apt install nginx) and then i moved the qcow to the /var/html/www folder.

proud-addition-21854

11/08/2022, 9:20 PM

I can test with nginx, yes

quaint-alarm-7893

11/08/2022, 9:20 PM

make sure you set perms too. i think i did chmod o+r {filename}

proud-addition-21854

11/08/2022, 9:20 PM

quaint-alarm-7893

11/08/2022, 10:21 PM

btw, w/ doing conversions like this, i've ran into issues later on if you do not have access the the qcow file. see this: https://github.com/harvester/harvester/issues/3016

basically, they are not included in backups, and on atleast v1.0.3 i havnt figured out how to export them yet as well

proud-addition-21854

11/09/2022, 9:04 PM

Thanks for letting us know.

Still have no luck even with nginx. I see the image completes downloading under /data/tmp on the pod backing-image-manager-xxxx-xxxx . Then throws the same error

I tested the same process on 1.0.2 and the image is successfully downloaded. On 1.1.0 on a different cluster (on nested vms), same issue.

quaint-alarm-7893

11/10/2022, 11:35 PM

😞 that's a bummer. i havnt moved to 1.1.0 yet, i'm waiting for the 1.1.1 patch for upgrades. cant help you much there. maybe open an issue. note you did it with nginx just fine on 1.0.3 though, so they know it's something w/ 1.1.0 specifically.

🆗 1

3 Views

#harvester Join Slack