I agree, I should be fired if that was the case. However, I often set things up and then it's up to other people to maintain it, and one thing I've learn't is to make it as idiot proof as possible...
I've had dozens of hard disks fail on me too, but I would have had 90% of thumb drives fail at some point, which is worse.. I pretty much hate thumb drives, and would never put anything critical on them.
By the sounds of it, at least some other people agree - I like the idea many people have said of using a card read as the ID - I agree that should be more robust (I've had far less of them fail than thumb drives..)
Yep, that's hard thing, and why it's probably been done with a thumb drive... If you make it based on -
- motherboard id - no good as if it goes (happened to me quite a few times) id will no longer match
- any of the hard disk ids - no good as they are going to probably change over time
- live internet activation - many of the boxes you'd want to use unRAID on might not be directly connected to the internet.
Yet you have to protect the IP somehow...
One of the things I've tried in the past is a crypto signed unique keyfile, with the customers name, address, email, phone number, and license number in plain text. While this won't stop a home user running a couple of copies, it tends to drastically reduce how many people they want to give it to, and does tend to stop most corporates from abusing IP.
On top of that, you then tie that key to support...
I agree that wouldn't be as good from a "IP' perspective as a usb key.
Maybe you could also add (to the above) a rolling hardware requirement, ie do the key for the initial hardware (disks and MB) and let it keep on working if any of that hardware still existed - and update the key automatically if new hardware was added/change if some of the original was still there (ie adding some disks, taking out some of the old ones). After all, if the MB and all the hard disks all go at once, there ain't going to be anything to run anyway... Yes, that would be open to 'replicating' piecemeal, but who is really going to bother with the hassle given the price of the software...
My final thought is feature based. ie make it able to run without the usb key - if previously one was present - with all the normal features (shares, number of disks, replace a dead disk etc etc) apart from that all the data is read only over the network . This would then make it easier to survive (and transfer to another solution) if it took a while (or wasn't possible) to get another key.