Snapshoting the VM prior to the copy offload breaks the progress. I assume this is because the disk has 0000 postfix and that breaks the task association, but not sure.
Beyond that, disk snapshot will break xcopy according to vSphere doc[1] and will defeat the purpose of the feature
[1] https://knowledge.broadcom.com/external/article/318941/frequently-asked-questions-for-vstorage.html

Always

1.create a plan with copy offload
2.watch the plan not completing the transfer disk phase
3.

not progressing

plan completes normally

culprit looks like commit f84a659bb on upstream/main