Incus: 7.3 Debian 13
Storage: lvmcluster (non-thin), QNAP iSCSI LUN, 4-node cluster
Summary
Copying an instance from the web UI creates the destination volume at the
pool’s default volume.size instead of matching the source. The copy then
fails with ENOSPC once dd passes 10GiB. The equivalent CLI command on the
same instance succeeds, so the mismatch is introduced by the UI.
UI (fails) — repeated on two different instances, both stopping at
exactly 640 records / 10GiB despite different throughput:
Create instance from copy: Failed to run: nice -n19 dd if=/dev/dm-19 \
of=/dev/dm-11 bs=16M conv=nocreat,sparse iflag=direct oflag=direct: \
exit status 1 (dd: error writing '/dev/dm-11': No space left on device
641+0 records in
640+0 records out
10741612544 bytes (11 GB, 10 GiB) copied, 55.56 s, 193 MB/s)
CLI (succeeds)
incus copy <source> <target> --target <cluster-member> --project default
Expected
The UI should either inherit the source volume size, or expose a root disk
size field in the copy dialog. Currently neither is possible, so any
instance with a root disk larger than the pool default cannot be copied
from the UI at all.