Pruning sender stuck with with old snapshot

[wojciechczyz]

I see that sender prunning job is stuck with snapshot from the past:

HDPool/LocalComputerDataFS/BackupsFS/HostsFS/local_backup/rpool ERROR: destroys failed @zrepl_20231114_215852_000: dataset is busy
HDPool/LocalComputerDataFS/BackupsFS/HostsFS/local_backup/rpool/ROOT ERROR: destroys failed @zrepl_20231114_215852_000: dataset is busy
HDPool/LocalComputerDataFS/BackupsFS/HostsFS/local_backup/rpool/ROOT/pve-1 ERROR: destroys failed @zrepl_20231114_215852_000: dataset is busy
HDPool/LocalComputerDataFS/BackupsFS/HostsFS/local_backup/rpool/data ERROR: destroys failed @zrepl_20231114_215852_000: dataset is busy

If I check the holds:
root@drassanes:~# zfs holds HDPool/LocalComputerDataFS/BackupsFS/HostsFS/local_backup/rpool@zrepl_20231114_215852_000
NAME TAG TIMESTAMP
HDPool/LocalComputerDataFS/BackupsFS/HostsFS/local_backup/rpool@zrepl_20231114_215852_000 zrepl_last_received_J_local_sink Tue Nov 14 22:58 2023

Why this happened and how to fix this?