[Pkg-sysvinit-devel] Bug#540311: show something to ponder while filesystem checking

[jidanni at jidanni.org]

Package: initscripts
Version: 2.87dsf-2
Severity: wishlist
File: /etc/init.d/checkfs.sh
X-debbugs-Cc: debian-devel at lists.debian.org

I have an idea. Every few days when booting we encounter a file system
check, which takes several minutes. We sit there, helpless,
unable to issue any command. All we can do is watch the progress marker.

Now being high powered Computer Pros (not me though), we ought to have
something smart put there on the screen for us to ponder as the check
proceeds. (No ads or fortune messages though please.)

Then it dawned on me, we sit there wondering "it says it is checking
/dev/hda6. What disk of mine is that that is taking so long? /home?
/usr?" OK, then tell us.

Sure, we can see on the screen that
it is /dev/hda6 etc. But our hands are tied, we cannot do
$ grep /dev/hda6 /etc/fstab
or any other command at this point, and even if we could,
/etc/fstab might have it listed as /dev/disk/by-id/... etc. and not
/dev/hda6. Therefore I'm proposing that just like you say
"Checking root file system", before checking it, you also say what the
other file systems you are checking refer to.

Here is the log of what we see.

# cat /var/log/boot
Fri Aug  7 12:53:17 2009: Checking root file system...fsck from util-linux-ng 2.16
Fri Aug  7 12:53:17 2009: /dev/hda1: clean, 16109/81280 files, 239546/325048 blocks
Fri Aug  7 12:53:17 2009: done.
Fri Aug  7 12:53:17 2009: Cleaning up ifupdown....
Fri Aug  7 12:53:17 2009: Loading kernel modules...done.
Fri Aug  7 12:53:17 2009: Checking file systems...fsck from util-linux-ng 2.16
Fri Aug  7 12:53:17 2009: /dev/hda6 has been mounted 35 times without being checked, check forced.
Fri Aug  7 12:53:17 2009: \002/dev/hda6: 44085/542720 files (14.7% non-contiguous), 3993396/4339408 blocks
Fri Aug  7 12:53:17 2009: /dev/hda7: clean, 12120/678912 files, 3955626/5428048 blocks (check in 5 mounts)
Fri Aug  7 12:53:17 2009: /dev/hda8: clean, 114339/814400 files, 587051/1627282 blocks (check in 3 mounts)
Fri Aug  7 12:53:17 2009: /dev/hda9: clean, 218558/950272 files, 1103243/1899442 blocks
Fri Aug  7 12:53:17 2009: /dev/hda10: clean, 1839/490560 files, 1298458/1959922 blocks

I notice a problem with these logs too, one needs to look at the kernel
times to realize that I stood there helplessly wondering 192-11=181
seconds. Hey, that's three whole minutes of Ph. D. time (not me
though)... let's be sure they are pondering something better than 'which
filesystem is /dev/hda6?'. And Ph. D.s don't keep hardcopy of previous
mount(8) output.

# cat /var/log/kern.log
Aug  7 12:53:21 jidanni2 kernel: [   12.450865] EXT3 FS on hda1, internal journal
Aug  7 12:53:21 jidanni2 kernel: [  192.161015] kjournald starting.  Commit interval 333 seconds
Aug  7 12:53:21 jidanni2 kernel: [  192.161323] EXT3 FS on hda6, internal journal

OK, now I can find the answer to the mystery:
# mount|grep /dev/hda6
/dev/hda6 on /home type ext3 (rw,commit=333)

Perhaps also mention some df(1) stats for us to think about while we are
sitting there. OK, I mean fdisk stats as it is not mounted yet.