A checking-only tool (or tools) that can be run in the background on a mounted volume would be more useful than an all-singing all-dancing automatic repair tool. One that could suggest running specific repair tools in the event of trouble would be more useful yet. After an ecosystem of checkers and repairers have got very mature, they would naturally be stitched together to be run automatically. Complaining doesn't speed up that work. Until it is ready, repair tools are best run on an image copy of the file system.
Posted Nov 3, 2011 6:35 UTC (Thu) by njs (guest, #40338)
[Link]
In the last thread, people claimed that checking-only tools have existed for some time (both online and offline, IIRC).
A btrfs update at LinuxCon Europe
Posted Nov 3, 2011 16:20 UTC (Thu) by iabervon (subscriber, #722)
[Link]
I think there needs to be one of them that's called "fsck.btrfs" so that it can be run by scripts between a detected bad event (e.g., kernel panic) and writing to the filesystem again.