[Liquidity X]

Ok, i just got Tech Tool 4, and I wanna defrag but says I can't because journaling is enabled, would it be bad to simply disable journaling, defrag and then re-enable journaling? TiA.

[Telusman]

I still don't understand peoples need to defrag a file system that doesn't need to be defragged. Unless your experiencing Severe performance degradation that i'd suggest you leave well alone. Defragging a journaled file system does not work because it corrupts the journal, i suggest you don't turn off Journaling and leave well enough alone.

-Telusman

Originally posted by Liquidity X:
Ok, i just got Tech Tool 4, and I wanna defrag but says I can't because journaling is enabled, would it be bad to simply disable journaling, defrag and then re-enable journaling? TiA.

[Cadaver]

...and, as I've read, Panther will auto-defragment files less than 20MB when they're accessed.

[Liquidity X]

yah well i push 20+ gig files around, and panther won't touch them. So is there a way to defrag and have it be safe?

[wataru]

Just turn off journaling, do your defragging, and turn it back on.

[pdot]

Originally posted by Liquidity X:
yah well i push 20+ gig files around, and panther won't touch them. So is there a way to defrag and have it be safe?

If you're going to be moving around such large files, you may want to give OS X or your workspace its on partition. That way, you won't have to worry (not for a long time, anway) about journaling and defragmentation of your OS partition.

[malvolio]

Originally posted by pdot:
If you're going to be moving around such large files, you may want to give OS X or your workspace its on partition. That way, you won't have to worry (not for a long time, anway) about journaling and defragmentation of your OS partition.

That's what I did. The system is on one partition, and all my major file-moving is done on a second partition.
BTW, I read somewhere that TTP 4 has a "Disable/enable journaling" tool. Do you see anything like that?

[Liquidity X]

Yah i disabled it in TT and then defrages, which took 14 hours, and then reenabled it.

[Millennium]

Disable Journaling, run the defrag, and re-enable Journaling. You cannot harm the system by doing this.

The only time your system would be at risk for corruption would be during the defrag operation itself. Journaling, however, would not save you in that case, because defragging would invalidate the journal anyway.

[jasong]

Yah i disabled it in TT and then defrages, which took 14 hours, and then reenabled it.

So now that you spent 14 hours where you couldn't use your computer while the defragmenter ran, how much faster is it? How long do you think it will take to get those 14 hours back?

Never understood the point of defragging the file system . . .

-- Jason

[Liquidity X]

its deff faster, and I let it run all night while I slept, so I lost not time at all.

[SMacTech]

Originally posted by jasong:
So now that you spent 14 hours where you couldn't use your computer while the defragmenter ran, how much faster is it? How long do you think it will take to get those 14 hours back?

Never understood the point of defragging the file system . . .

-- Jason

I guess you never have used a drive where the files are heavily fragmented. Especially one where you are writing large files, as Liquidity is. Maybe his computer de-fragged while he was sleeping, oh wait he did do it then.
If files didn't need to be de-fragmented, why would Panther have it built-in now, albeit for files <20mb?
The point is that it does add a level of performance increase, maybe in some cases it is only 1%, others may be substantially more, depending on many things.

[jasong]

I am not arguing that defragging helps, I just question the value of it. A process that takes 14 hours to run and only delivers a 1 - 10% speed increase gives you very diminished returns (i.e. your Photoshop filter that takes 100 seconds to run, now runs in, at best, 90 seconds). The new HFS+ method for de-fragmentation takes no time to run, so anything that transparent with any speed gain is worth it - good move on Apple's part.

Anyway, you are right that I don't deal with large files so much anymore. I am a web developer and virtually everything I work with is tiny. I do a lot of iMovie work, but I tend to only work on one project at a time, and I have a separate disk that holds all my iMovie files for the current project. I reformat the disk prior to each project, so I doubt this gets very fragmented.

And, I apologize if my tone was derogatory in my original post, my intent was only to question the value of de-fragmentation, and I think it came out a little harsh.

-- Jason

[SMacTech]

Originally posted by jasong:


And, I apologize if my tone was derogatory in my original post, my intent was only to question the value of de-fragmentation, and I think it came out a little harsh.

-- Jason

I didn't take it that way, but anything to improve the performance of my Mac is welcomed. TechTool's optimizer always took a very long time. I use Norton SpeedDisk infrequently, and it usually can do it's thing during lunch on my OS partition. My data drives are another story.

[Dale Sorel]

Originally posted by Liquidity X:
Yah i disabled it in TT and then defrages, which took 14 hours, and then reenabled it.

[zigzag]

One perspective on defragging: http://db.tidbits.com/getbits.acgi?tbart=07254

[davidb224]

Originally posted by Cadaver:
...and, as I've read, Panther will auto-defragment files less than 20MB when they're accessed.

Actually there two separate file optimizations going on in Panther.

The first one, as you stated, is Automatic File Defragmentation. When a file is opened, if it is highly fragmented (ie. 8+ fragments) and it is under 20MB in size, it is defragmented. This works by just moving the file to a new location. This only happens on Journaled HFS+ volumes.

The second is called "Adaptive Hot File Clustering". Over a period of 60 hours, the File System keeps track of which files are read most frequently (to be a hot-file candidate, a file must be less than 10MB and never written to).

At the end of the period, the "hottest" files (ie. the files that have been read the most times) are moved to the "hotband" on the disk (which is that part of the disk which is particularly fast given the physical characteristics of the disk).

The size of the "hotband" depends on the size of the disk (ie. 5MB of hotband space for each GB of disk). "Cold" files that were in the hotband will be moved out of the hotband to make room for the hot files. As a side effect of being moved into the hotband, the hot files are also defragmented.

Currently, Adaptive Hot File Clustering only works on the boot volume, and only for Journaled HFS+ volumes that are more than 10GB.

So, AHFC optimizes the location on the hard drive of the files you read the most and in the process defrags those same files "on the fly".

[OmniX]

i have heard and read (and seen the Darwin source) about Panther's auto <20MB defragging, but Adaptive Hot File Clustering in Mac OS X is news to me.
Is this a new feature as of Panther? Can anyone else confirm its existence (preferably citing official Apple documents or source)?

[mrmister]

It is new in Panther, though I don't have any references for it handy. Note that defragging your whole drive undoes this, so in a way you'll be taking a step back while you take a step forward.

Defragging isn't worthless--it just isn't anywhere near as useful as it used to be.

[tkmd]

I use cocktail to clean out sys/vm swap file /user cache's. Will deleting these caches erase the log kept on the most used files for AHFS?

[wataru]

Originally posted by tkmd:
I use cocktail to clean out sys/vm swap file /user cache's. Will deleting these caches erase the log kept on the most used files for AHFS?

That information is probably well hidden. I don't think it's something you could delete even if you wanted to.

[ngrundy]

wouldn't it be somewhat easier and probably quicker to copy the data off onto a backup/external drive. Format your work drive and copy the data back?

As a formated drive has no data on it the 'fragmented' files would stream back onto the drive in one piece.

[chabig]

That information is probably well hidden. I don't think it's something you could delete even if you wanted to.

It's not very hidden. It's at /.hotfiles.btree

So deleting cache files won't affect the hot file clustering.

Chris

[digital_dreamer]

I could never stand defragging with Drive 10, and would stop it well before finishing because it�s way too slow. TTP 4 seems to use the same technique - S-L-O-W.

Although I use Diskwarrior for fixing directory problems, Norton�s Speed Disk is the tool to use for defragging.

I just used Speed Disk on a 80GB HD that was 90% full on OS X 10.1 Server last weekend. It defragged some 300,000 files in 4 hours. Some of the files were fragmented in over 1,000 pieces (one was >1,800).

I wouldn�t install any of the crap from Norton�s SystemWorks (it doesn�t even let you modify the installation ). But, Speed Disk is the bomb when it comes to physically reorganizing your files on the HD, IMHO.

regards,
MAJ

[Arkham_c]

Another point on defragmenting -- if you have enough disk space, you will not get significant fragmentation to start with. When HFS+ allocates space, it attempts to minimize fragmentation by writing to the largest contiguous block. If you always have plenty of space, you won't get fragmented to start with.

[SMacTech]

One of my Macs with a 40gb drive that I used to install the Panther betas on had very little fragmentation as reported by speed disk. It wasn't optimized, as files were spread all over the disk. However, the number of non-contiguous files was something around 400 out of 200,000 files with 10gb of free space. This was after about 7 or 8 installs of panther. Not too shabby, I think.
So as Arkham_c pointed out, and I concur, a lot of this is impacted by free space on the drive.

[orry_main]

Originally posted by SMacTech:
One of my Macs with a 40gb drive that I used to install the Panther betas on had very little fragmentation as reported by speed disk. It wasn't optimized, as files were spread all over the disk. However, the number of non-contiguous files was something around 400 out of 200,000 files with 10gb of free space. This was after about 7 or 8 installs of panther. Not too shabby, I think.
So as Arkham_c pointed out, and I concur, a lot of this is impacted by free space on the drive.

Is there any easy way to find out just how fragmented a drive is? How did you find out that ~400 files of 200k were fragmented, for example?

[chabig]

Is there any easy way to find out just how fragmented a drive is?

He said the fragmentation was reported by Speed Disk. That's a Norton product.