Yeah ZFS is shit especially since Oracle purchased Sun and fired everyone.
Too bad Butters FS isn't ready (btrfs)
I wish there were a new filesystem, or a new OS that did away with this ancient way of filing data. It's almost 2020 and we're still using 1990's technology. WTF?
Presumably they're referring to Datto's recovery process (i.e. recovering automatically from secondary or tertiary backups), as opposed to just the ZFS mirror.
ZFS volumes can also be backed by files. In this case, the volumes can be corrupted by opening the backing file in a text editor, and perform a search/replace. This is simpler but of course less interesting :-)
Being able to back volumes by files is an amazing feature for testing. I've lost count of the number of times I've done this to prove some documentation works before applying it to the actual storage array in question.
8 comments
[ 3.8 ms ] story [ 21.1 ms ] threadhttps://www.kernel.org/doc/Documentation/device-mapper/
In this case, the recovery (repair) of the system (that is, to bring it to a clean state) simply consists in deleting the file (see https://docs.oracle.com/cd/E26505_01/html/E37384/gbbbc.html#...).
If there was an intact copy of the file available in the mirror, then one doesn't need to do anything, as ZFS is self healing: https://blogs.oracle.com/timc/demonstrating-zfs-self-healing.
Another interesting article I've mentioned in another post, is a demonstration of ZFS' self-healing capabilities: https://blogs.oracle.com/timc/demonstrating-zfs-self-healing.