Failing image uploads through sunstone to ceph image datastore

Product Support Operations
1

Please, describe the problem here and provide additional information below (if applicable) …

Versions of the related components and OS (frontend, hypervisors, VMs): HCI cluster of 4 nodes with OpenNebula 7.0.1 KVM hosts and ceph 19.2.3 on Ubuntu

Steps to reproduce: If I try to create image by uploading an .iso from my client computer to the ceph image datastore it fails with the following error:

Tue Dec 2 08:59:32 2025 : Error copying image in the datastore: 0+0 records in0+0 records out0 bytes copied, 2.1338e-05 s, 0.0 kB/sError dumping /var/tmp/1de98308e9612f88bdc98dc8f3597d50 to localhost:/var/tmp/161d82587eb111dbe4987192aea6a237

As you can see from the error the bridge_list is currently set to localhost, but has earlier been set to the IP of the frontend or one of the KVM hosts. This seems to have no influence on this error.

If I use scp to transfer the .iso to the frontend /var/tmp directory, and then create an image from this path/file the image is created successfully, so the transfere to the ceph image datastore seems to be possible.
When I had the bridge_list set to one of the kvm hosts I thought that the problem was the transfere from the frontend to the kvm host by passwordless ssh, but when I tried manually this was not the problem, and thus I also have enough space on the disk for /var/tmp.

I’m also able to store VM templates in the ceph image store and run VMs in the ceph system store, so I think these datastores should be correct setup. I can also download apps/templates from the markedplace into the ceph image datastore without issue.

Does anyone have a clue about what the problem might be?

Current results: Error copying image in the datastore: 0+0 records in0+0 records out0 bytes copied, 2.1338e-05 s, 0.0 kB/sError dumping /var/tmp/1de98308e9612f88bdc98dc8f3597d50 to localhost:/var/tmp/161d82587eb111dbe4987192aea6a237

Expected results: Success

2

Hi,

Thanks for the detailed report. We were able to reproduce this behavior and can confirm that it is caused by a regression in the image-upload workflow in OpenNebula 7.0.1. Although the original GitHub issue was about uploaded files not being deleted from /var/tmp, the same code path also introduced the upload failure you’re seeing.

Error copying image in the datastore:  
0+0 records in  
0+0 records out  
0 bytes copied ...

Relevant issue is here (the fix is being handled under this case).

The problematic cleanup logic has already been fixed in the development branch, which will become OpenNebula 7.2 CE.

If you’d like to receive the fix right away, you can build OpenNebula from the current master branch, where the corrected behavior is already included.

3

Thanks for looking into this. The most important thing for me currently is to know that it is’nt just a simple configuration error on my side. I have a workaround until the fix is released in 7.2.