To distill down some cold hard facts come to a fairly inescapable conclusion.
1. ZFS doesn't have a "fsck" for many of the same reasons that ZFS has no recovery tools. When properly managed ZFS should be impossible to corrupt without seriously bad hardware and/or extremely poor administration. ZFS's structures, because of how they are engineered, are one of those things that "work or don't work". If you know they "don't work" because the pool is unmountable, how do you handle the recovery?
2. When a pool becomes corrupted you are often left with a situation where you have to trust data on diskA or data on diskB since they don't agree. That's a decision that is really hard (if not impossible) to accurately master in code.
3. ZFS is marketed for enterprises. You know, companies that can add an extra zero to cost of the server and not care. These are also the same places where backups are done religiously and all the time and recovery isn't something that a company is going to need, let alone want. Restoring from backups is going to be faster and better for a whole list of reasons
4. Feature flags. These are a mess to deal with as every OS is implementing them differently. While the actual structures on-disk should be the same be it Illuminos or FreeBSD, how they "get there" is often quite different. Sometimes, these peculiarities are important for recovery.
Put all this together and you have a perfect storm of reasons why ZFS recovery is very difficult to "just do" and isn't particularly profitable. No money and low demand means no product will come to market.
Yes... yes... I know you're going to say that ZFS is moving into the home.. blah blah blah. Ok, but look at the reality of the situation. Like 1/2 of all PBIs that exist on FreeNAS are for pirating stuff. Even if I came out with some stellar ZFS recovery tool, how many people would actually pay the market value for my product? I'd expect a pricetag between $1000 and $2000 for "small home server" license to be a fair price, but people would just assume torrent it and I'm out the money. So that "no money and low demand" just went from extremely poor to almost zero. Nearly 100% of the people that lose their pools lose them because of their own choices of hardware, RAIDZ type, etc. They always make those choices because of at least one of the two main reasons.. money or knowledge. If it's money, they sure as heck aren't going to buy my tool as it probably cost them more than their entire server. They're going to run to piratebay or whatever the popular torrent website is and rip it off. Poof, even less of a reason for a product to actually come to market.
I'm really not sure why someone wants to have a discussion on this topic. The market forces and the realities of ZFS are the same as they were 5 years ago, and I bet that 5 years from now not much will have changed. ZFS might be much more used than it is now, but it's never going to be a "home product". Just look at Windows. They implemented their own crappy ZFS "wanna-be" into Server 2012 R2. Apple considered ZFS for it's products, then abandoned it. ZFS is a professional product and will always be. In fact, year over year its becoming more difficult to keep up with because more feature flags and such are being added. So the problems ZFS has are only getting more complex.
