0

I have an old Titanium that has bad sections on the drive. This isn't my main system, I use it to check e-mail, listen to a voice mail server, etc. It's basically a semi-retired, but still useful. Some weeks ago I while using it I started getting spinning beach balls periodically. I thought the systems memory was all used up but Activity Monitor told me otherwise. The system was also making some noises when it would slow down.

After doing some research I came across this article:

http://www.scsc-online.com/Bad%20Hard%20Drive%20Signs%20and%20Symptoms.html

The noise they refer to as a "chug-a-chug-a" noise was exactly what I was hearing, so I went ahead and bought their drive testing tool Scannerz. I ran the program and the drive scanned perfectly until about the last 3GB of the drive. When entering this realm, Scannerz, which had been moving along very speedily suddenly slows down. Irregularity counts, which were at 0 suddenly start increasing at about 1 per second. Then Scannerz slows down even futher, then the error increment goes up. This happened through the last 3GB of the test.

I put Scannerz into probing mode and did a test of the last 3GB of the drive to determine how severe the damage was. After about 15 min of testing I cancelled the test because it was quite evident that the damage in these areas is quite extensive.

This are really a 2 questions I have:

Part 1: The drive in this system was replaced from it's original 40GB to an 80GB drive some time ago. The 40GB drive wasn't big enough at the time to hold all my stuff. I ended up putting it in an iBook. I would guess around 2006. The 80GB drive only has about 50GB used on it.

Here's question 1: If this thing is using ONLY 50GB of space, why is it causing me problems in the 77-80GB range? I thought the operating system wrote things sequentially keeping everything nice and orderly.

Part 2: This system is old and I don't want to replace the drive. IDE hard drives are now commanding a premium price (twice what SATA drives cost) because a good, new one is hard to find. A new IDE drive would likely cost more than computers like this are selling for on ebay. This system is so old that I know that one of these days I'll open it up and a puff of smoke will come out of it, the LCD backlight will be dead, or it just won't turn on. This will happen, probably within the next year. The thing is already 11 years old.

With that said, what I'm thinking about doing is partitioning the drive to EXCLUDE the last 3GB. I figured I could split the partition up into one 77GB volume and a 3 GB volume, then delete the 3GB volume.

If I did this, would it completely eliminate the bad region of the drive from being accessed by the OS, and thus put an end to the spinning beach ball problems. I think it will. Can anyone confirm or dispute this?

3 Answers 3

0

How a drive reads and writes blocks of data varies widely with operating systems and media types. On MacOS with a traditional hard disk, when a file is deleted, it isn't necessarily deleted, but rather the blocks get marked as "available." HFS+ will try to collect as many contiguous blocks that are either free or marked as being available as possible during a write operation. This cuts down on defragmentation. Some drive utilities have "recovery" mode, and when they attempt to recover deleted data, what they're really doing as looking at all the blocks marked as being available that contain data.

SSDs, FWIW are different because unlike an HD that can mark the block as available and overwrite it at any time, an SSD block needs to be cleared or erased before it can be written to again. The erase is incredibly slow with an SSD, so SSD makers have implemented "TRIM Support" on drives, which clears available blocks as a background process.

The reason that you're running into bad sectors at the tail end of the drive is because when the OS attempts to find contiguous free blocks, it attempts to use up all those that have never been used at all first, THEN it starts overwriting the "available" blocks that used to contain data. As a result eventually a file will end up being written to the end of the drive, even though there may be plenty of apparently free space of the drive.

Regarding reducing your partition size and blocking out the bad area of the drive, that might work, but even HFS+ uses some region of the tail end of the drive. Apple has a document out that describes how the file system is laid out that you can review in the following link:

http://developer.apple.com/legacy/mac/library/#technotes/tn/tn1150.html

I'm not sure that's up to date, but if you go through it you'll notice that it places some information at the tail end of the volume, so you might want to make absolutely certain that your volume ends before the bad sectors begin.

To find the location where to terminate your shrunk partition, it should be in the Scannerz log files for your tests. Make sure there are NO irregularities in that area because they're probably weak sectors, damaged by the head crash. You could try to re-map the bad sectors to spare sectors which was implied in Brad Patton's answer, but odds are that if Scannerz was finding tons of them, there are likely to many to be remapped. I would speculate that by shrinking the size of the volume, writes will never occur on that bad region of the drive, and thus the spare sectors might remain available (assuming there are any left) should a bad sector crop up in the good section.

For kicks I did a search for used Titaniums and new IDE drives. The typical ebay price for a Titanium can range from about $30US to $120US (in excellent condition), whereas a new, unused IDE of about 120G was running over $100, so I can understand why you want to consider a more cost sensitive approach.

0

I'm not really answering my own question, I'm just going to tell you what I did.

I didn't want to attempt to re-map the bad blocks because I figured the damage at the end of the drive was so extensive it would eat up all the spare sectors, if there are even any left.

I did a back up of the drive, then using the log files from Scannerz found the location of the blocks on the drive where the damage began. Using this data I split the drive into two volumes, one which is still big and occupies most of the drive, and the other at the tail end of the drive that has the bad data. I moved the start point for the second, bad volume about 500MB ahead of its starting point just to be safe. I formatted both partitions NORMALLY - no zeroing - and then DELETED the partition containing the bad regions of the drive.

I retested with Scannerz. If I select my newly created good volume, it passes tests with flying colors. If I select to test the entire primary drive which will be the entire drive including the deleted volume the problems are still there - that's just to clarify that the problems haven't magically disappeared, they're just not visible anymore.

I re-installed the OS and my data and now everything's running fine.

How long will it be good? Who knows. The drive's ancient and I'm surprised it's worked as long as it has. I know one of these day's I'll go to turn this old friend on and the drive will be making grinding noises, the light on the display will be gone, or maybe it just won't turn on at all. I know this old friend is old and it will quit on me some day...

.....BUT NOT TODAY!!!! :-)

0

While initially data is written to the drive in neat orderly ways it doesn't stay that way for very long. As files get created and deleted (not just your data but OS files like swap) the blocks of data that contained those files get released. As new files get created the OS will reuse those blocks but will also use new blocks at the end of the list. BTW this is one of the reasons that data on hard drives get fragmented. HTS+ has a number of ways to reduce this fragmentation (see here) but it will still happen to some degree.

So the question isn't really which order does data get written but can the whole hard drive platter be accessed for reading and writing data. The answer is yes.

If you were certain all of the bad sectors of the drive were at the end you could potentially wall them off with a separate partition. Given the age of this drive I'm not sure I would agree.

I would recommend reading this SuperUser question Fix bad blocks on Mac hard disk and following some of the suggestions on using tools to make the bad blocks so that the OS doesn't attempt to use them.

Oh and have back-ups because as you note this thing could die any day.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .