Check Disk State

When a server comes back online after a crash or an unclean shutdown, one of the first things you want to confirm is whether each ext3 or ext4 filesystem is in a clean state. The commands below walk every mounted partition, pull its label, and print the Filesystem state line from tune2fs so you can spot dirty or error flags without digging through logs.

ext3 only

grep ext3 /proc/mounts | gawk '{ print $1 }' | sort | uniq | while read part; do
  echo "$part - $(e2label $part) - $(tune2fs -l $part | grep state)"
done | column -t

ext3 and ext4

egrep '(ext3|ext4)' /proc/mounts | gawk '{ print $1 }' | sort | uniq | while read part; do
  echo "$part - $(e2label $part) - $(tune2fs -l $part | grep state)"
done | column -t

What to look for

• Filesystem state: clean – the partition was unmounted properly.
• Filesystem state: error – the kernel detected a problem; run fsck on that device before trusting the data.
• Filesystem state: dirty – the filesystem was not cleanly unmounted. A read-only remount followed by fsck will usually resolve it.

The column -t at the end lines everything up into a tidy table, which makes it easy to scan a box with a dozen or more disks.

• Technology
• How-To

• SysAdmin
• Bash