I ran into this error recently when I partially wiped a member of a zfs pool accidentally. On a 6TB member disk, I wiped roughly 200GB before realizing I started wiping the wrong disk. ZFS did what it does and marked it as faulted after finding a bunch of checksum errors as you would expect (though it did attempt to resilver them).
On next boot, that drive faulted stating the label was missing. No big deal I thought. I just ran 'parted' and did a 'mklabel gpt' on it. Interestingly enough, the drive sprung to life again and began resilvering. However it eventually faulted again for the same checksum issue.
Scrubbing always failed with that drive being faulted shortly afterwards. I figured I would replace it on next boot later in the day as this was a non-production system. Wouldn't you know it, subsequent boots would fail with this 'alloc magic' error. At first I did not equate this problem to the drive I managed to convert into magnetic confetti with my antics. After disconnecting this drive (1 of around 60) and ONLY this drive, the error disappeared and I was able to replace the drive successfully.
So yeah, don't partially wipe stuff and think zfs will make it healthy. But I will praise zfs for its resiliency for the rest of the pool and its never-give-up attitude with trying to fix broken things.