Howdy! So I've got a server (ubuntu 11.10) that's operating on a single board computer in a somewhat hostile environment and occasionally just gets the power removed, rather than having a nice shutdown run. It doesn't normally have a monitor / keyboard hooked up. When it's rebooted, the system sometimes detects that there are some problems with the drive and fsck should be run. Unfortunately it just sits there waiting for a user to come by and hit 'f' to 'fix' the errors on the drive. This requires someone to go to the physical location of the server with a power supply, monitor, keyboard, and plug in and hit 'f' a few times and then the server is back up and running and I can ssh in and check on stuff.
Is there a way to force fsck to just always automagically hit 'yes' / 'fix' when this unusual situation appears? I looked in tune2fs and fstab, but couldn't find anything about the OS behavior when errors are detected in the file system. I'd rather have some drive/data corruption on rare occasions (and potentially risk the server) than to have to physically be present at the server on those rare occasions.
I am attacking the other side of the problem (preventing the power outages) as well, but if there's a software fix, that's always nice.

Thanks!

That's in the options that are passed to e2fsck

Try e2fsck -f -C -y

HI,

Open /etc/fstab and "1" at the end of the partition line. It will force the partition to do file check on every time the system boots

/dev/sda7 /partition_name ext3 defaults 0 1

Here's an illustration of the problem:

http://i.imgur.com/wVjBH.jpg

When the system boots, it runs fsck and finds errors. The problem is that it insists that a human come over and hit 'f' to fix the problems.
Many have said that there's a script that runs fsck on boot, and that I just need to edit that script, but I have not found it. It doesn't seem to reside in /etc/. Here's the results of "grep fsck /etc -r"


./init/mountall.conf: [ -f /forcefsck ] && force_fsck="--force-fsck"
./init/mountall.conf: [ "$FSCKFIX" = "yes" ] && fsck_fix="--fsck-fix"
./init/mountall.conf: exec mountall --daemon $force_fsck $fsck_fix
./init/mountall.conf: rm -f /forcefsck 2>dev/null || true
./bash_completion.d/lvm: --size --nofsck --resizefs --test --verbose --version' -- "$cur" ) )
./bash_completion.d/lvm: --extents --size --nofsck --resizefs \
./bash_completion.d/lvm: --extents --size --nofsck --resizefs \
./bash_completion.d/git: fsck-objects) : plumbing;;
./bash_completion.d/git:_git_fsck ()
./bash_completion.d/git: receive.fsckObjects
./update-motd.d/98-fsck-at-reboot:if [ -x /usr/lib/update-notifier/update-motd-fsck-at-reboot ]; then
./update-motd.d/98-fsck-at-reboot: exec /usr/lib/update-notifier/update-motd-fsck-at-reboot

Any ideas? What part of the OS actually calls fsck under these conditions?

Something like this (from slackware) in the rc scripts will be in charge

grep -C3 -r fsck /etc/rc.d/* usually finds it

Use the below command to find out what partition it belongs to
df -T /etc/fstab