Ext4 self errors after some time on CentOS 5.8 64-bit

Question

I get such warnings in dmesg:

EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619328(bit 15808 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619329(bit 15809 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619330(bit 15810 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619331(bit 15811 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619332(bit 15812 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619333(bit 15813 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619334(bit 15814 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619335(bit 15815 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619336(bit 15816 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619337(bit 15817 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619338(bit 15818 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619339(bit 15819 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619340(bit 15820 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619341(bit 15821 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619342(bit 15822 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619343(bit 15823 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619344(bit 15824 in group
EXT4-fs error (device md2): mb_free_blocks: double-free of inode 0's block 602619767(bit 16247 in group
JBD: Spotted dirty metadata buffer (dev = md2, blocknr = 0). There's a risk of filesystem corruption in
EXT4-fs error (device md2): ext4_mb_generate_buddy: EXT4-fs: group 18390: 7176 blocks in bitmap, 7211 in
JBD: Spotted dirty metadata buffer (dev = md2, blocknr = 0). There's a risk of filesystem corruption in

This is just happening after some time. The server wasn't rebooted. All disks are fine. I use Software RAID1 setup on two 3TB drives. I don't see any issues, instability or something else.

One time I've ran e4fsck during the night. It corrected all tje errors and then it started to happen again, but server wasn't rebooted at all.

Is there something to worry about ?

Kernel:

2.6.18-308.4.1.el5 #1 SMP Tue Apr 17 17:08:00 EDT 2012 x86_64 x86_64 x86_64 GNU/Linux

Answer 1

Yes, it's something to worry about. It means either something is corrupting your filesystem, or something else is corrupting your filesystem.

Prime suspect will be your disks themselves. Check them with smartctl:

smartctl -a /dev/sda
smartctl -a /dev/sdb

Next suspect will be the drive controller and cabling.

Answer 2

I had to upgrade to kernel 3.X.X where ext4 works just fine ;)