After having converted most of my filesystems to ext4, I do have some problems with one of the filesystems. Quite often the following messages appear in the dmesg output (or logfiles):

EXT4-fs (md5): error count: 26 EXT4-fs (md5): initial error at 1295452467: ext4_lookup:1043: inode 45679232 EXT4-fs (md5): last error at 1295453545: ext4_lookup:1043: inode 61022838

Not looking to closely at the message, I assumed it was a filesystem problem. Being quite allergic to losing data, I stopped everything and did a fsck, which turned up nothing.

After looking a bit more closely at the problem it turns out the ext4 filesystem has different ways of logging information and has a memory for errors. The type of message can be identified from the beginning of the message.

  • EXT4-fs (device): This is an informational message, nothing to worry about.
  • EXT4-fs warning (device): This is a warning message.
  • EXT4-fs error (device): This is an error message.

So it tuns out the message I saw were just informational. According to some posts by Ted Tso the message is repeated daily if there has been a problem with your filesystem, just to remind you. Normally this information will be reset when the filesystem has been checked, but this only happens from e2fsck version 1.41.14 (I am running 1.41.9).

So what looked like potential data loss turns out to be a quite useful feature of the ext4 filesystem, as long as you know what you’re looking at :)