ZFS snapshots are not deleteted passed their expiration date

[newhost]

I have a production machine which is set to snapshot the ZFS volume once per day with a retention time of one week. The problem is that the snapshots that are older than a week are not deleted.
Without manual intervention one or more dataset can fill up and beside the fact that no more data can be stored, if I do any modification in the iSCSI configuration the daemon will not restart, thus bringing down the service for all datasets. You can't imagine what happens then. There's no warning about the reason for which the iSCSI daemon does not restart, only a message saying that the service can not start. I've accidentally found out the reason and since then I am constantly keeping a close eye on the snapshots and the datasets available space.

[william]

I believe this is a duplicated of #806