Running "fsck" in the Background

Traditionally, recovery of a BSD fast filesystem after an uncontrolled system crash such as a power failure or a system panic required the use of the filesystem checking program, "fsck". Because the filesystem cannot be used while it is being checked by "fsck", a large server may experience unacceptably long periods of unavailability after a crash.

Rather than write a new version of "fsck" that can run on an active filesystem, I have added the ability to take a snapshot of a filesystem partition to create a quiescent filesystem on which a slightly modified version of the traditional "fsck" can run.

A key feature of these snapshots is that they usually require filesystem write activity to be suspended for less than one second. The suspension time is independent of the size of the filesystem. To reduce the number and types of corruption, soft updates were added to ensure that the only filesystem inconsistencies are lost resources. With these two additions it is now possible to bring the system up immediately after a crash and then run checks to reclaim the lost resources on the active filesystems.

Background "fsck" runs by taking a snapshot and then running its traditional first four passes to calculate the correct bitmaps for the allocations in the filesystem snapshot. From these bitmaps, "fsck" finds any lost resources and invokes special system calls to reclaim them in the underlying active filesystem.