Frequent 'Filesystem Corrupt', Invalid File count

Question

I have been getting this message every a couple of days in my MBP or iMac running 10.6.2. It happens/visible only when I runs Disk Utility and Verify Disk. Of course, this brings unsettling point to me when it comes now and then. Not to mention I have to boot from DVD to run disk utility to fix this
Error Message: Invalid file count (It should be xxxxx instead of xxx+1) AND
Error Message: Invalid volume directory count (It should be xxxxx instead of xxxx-1)
Any idea what causes this and what's the impact i.e. do I lose any files?
Thanks!!

Answer 1

Leonard Ong wrote:
Hi Tim,
By default, it is HFS+ Journaled (non-case sensitive). So I have also the same understanding that journalled filesystem should be robust as you wrote it. But the fact that this comes now and then, It created some doubts.
Is this a harmless error message or should I be worried about it. That basically my questions. I do repair when I see this.
It can't be because of disk is failing for a number of reason:
1) I have the same behavior both in my iMac and MBP
2) Both HDD has been replaced with new one recently
3) Both HDD has passed through surface scan from TechTool Deluxe 5.0.6, so no bad blocks
Still a puzzle...
Are you sure that the errors are showing up on both computers when performing either a "Verify Disk" or "Repair Disk" and NOT just a "Verify Disk Permissions" or "Repair Disk Permissions"? Permissions errors are mostly harmless. Filesystem errors (the sort you see when you ask for "Verify Disk" or "Repair Disk" in Disk Utility are not harmless and you should be worried about those.
If so, then either both drives are failing or your drives are being exposed to a strong enough magnetic field to damage their data (and that would have be pretty strong -- a common refrigerator magnet wouldn't do the trick.) In all of my years of working with journaled filesystems (or operating systems with "filesystem logging" which is the same thing) I have NEVER seen a journaled volume have filesystem errors unless it was physically failing.
Journaling will prevent filesystem errors if it's enabled for pretty much every cause except for a physically defective or damaged disk. Here's the short version of how journaling works...
Journaling (aka "filesystem logging") is intended to prevent filesystem corruption... and it's pretty rock-solid.
1) A "journal" (or "log") area is added.
2) A hard rule is established that, under no circumstances may the OS be performing writes to both the journal and the regular filesystem at the same time. One must be flushed before it can work on the other and vice versa.
3) When a write operation is about to happen, the OS writes information to the journal indicating which blocks on the filesystem are about to be changed and what it's about to change. This information is flushed to disk.
4) The OS can then perform the write on the actual filesystem areas it intended to change. These must then be flushed to disk.
5) The journal can now be cleared.
That's pretty much it (simplified). But think about what this does. Normally if something interrupts a write operation (something crashes, a kernel panic, a power loss, etc.) there is something that will still have integrity because the OS will not write to both the journal and the regular filesystem at the same time.
If a failure occurs while writing to the filesystem it's no big deal. When the system boots back up, it notices the journal wasn't cleared so it "replays" the journal. This re-writes any blocks that might only have been partially written thus clearing any chance of corruption on the volume.
If a failure occurs while writing to the journal it's still no big deal. When the system boots back up it notices a corrupt journal, but that means the regular filesystem itself wasn't being touched so the filesystem is guaranteed to not have errors on it (although it does mean the last file you were about to save probably didn't get saved -- and that file wouldn't have written successfully even if journaling were off. At least this way there is no filesystem corruption.)
While it sounds pretty solid (and it is) there's still a potential for corruption, but it's rare. If a drive swings the access arms to a specific track and starts writing data, the assumption is that the write head is above the correct track. If the head is "sloppy" due to wear, then the head may be slightly mis-aligned and bleeding writes into the adjacent track. This will cause filesystem corruption even with journaling enabled.
If these drives came with the computers then the only warranty would be via Apple. If you purchased these drives on your own then they probably came with a pretty good manufacturer warranty -- most drives I've purchased in recent years had at least a 3 year warranty.