Atte Andr? Jensen <atte.jensen(a)gmail.com>om>, on Wed May 24, 2006 [09:08:33 PM] said: Hi; In general, fsck is invoked on all filesystems before mounting. If the filesystem is marked dirty, then fsck does something. (for the journaling filesystems, thats usually just replaying the journal) However for ext2/3 filesystems, there are two ways a full fsck will occur, that is if the maximal mount count has been reached, or if the maximal time since the last full fsck has passed. You can set these values via 'tune2fs' -i and -c options. On a laptop especially, turning off the maximal mount count is helpful, and if you dont want any unexpected fsckage, then disable the maximal time too. eg.: tune2fs -c0 -i0 /dev/hda1 Another method of controling long fsck times, is to stagger these settings, so that all the filesystems dont get fscked at the same time, but thats not helpful if you just have one large filesystem for the most part. The reason these options exist, is because there are more ways than just graceless shutdown (leaving the fs marked 'dirty') that can corrupt a filesystem, and its a good idea to force a check every once in a while. I think most distributions will have a flag you can set, that will either force an unconditional fsck next boot, or skip all fsckage. eg. under gentoo, creating a file /fastboot skips fscking, and creating a file /forcefsck does just that. These are one-shot flags and are removed after the boot. (and obviously you have to think to set them before shutting down, in anticipation of the next boot;) Interrupting (^C) a 'benign' fsck shouldnt hurt you, but if fsck is actually doing any repairs, I wouldnt touch it. Paul set(a)pobox.com