record fragmentation stops recording!

[drumpriest]

Headline, headline....

We recorded the drum tracks for our CD this weekend. After a day off, we went back and listened to the tracks to ensure that we were satisfied with the results. There were 2 sections to redo.... 1 because of excessive snare buzz, and 1 because I was slightly off. I went back to punch in and retrack those sections, but couldn't get my system to do it. It kept dropping out due to disk read speed.

After some frustration, we gave up for the night, and I decided to defrag the drive. Now it's a 250 gig hard drive, with 21 gigs used, or 8% space used. It's 99.2% fragmented.

Unfreaking believable! Windows must have had a fun time allocating those files we recorded to get it that messed up. I am blown away by it.

So friday we punch in for repair tracking, and it should be done....

I have a small log of the recording that I'll post at that time, if anyone is interested.

[TheRic]

.... Now it's a 250 gig hard drive, with 21 gigs used, or 8% space used. It's 99.2% fragmented. .

WTF Are you sure you have those number correct?!! Did you fill the hard drive at some point, then deleted a bunch of stuff?!! If not there is something wrong! I have never heard of such a thing. I have seen disks with 99% defrag before, but not one with that much space free!!

[bmyers]

Yeah really. That does sound strange. You must have had it much more full at one time.

Do post a sound byte though. I'm interested..

Bill

[drumpriest]

No, it was never even remotely close to full. The trouble is in the way that these drives are accessed while doing multi-track recording and using several takes. A lot of deleting and putting things back. Also, writing 8 large tracks while reading back xxx large tracks off of the same drive setup at the same time. One would think that Sonar would have a more sophisticated algorithm for determining file placement on the drive, but it appears to just use win32 functions. This is a problem as the file system cannot predict that this new file will be hundreds of megs in size when the file is created.

[crokett]

Keith,
Not sure what your requirements are but you might want to look at a RAIDed set of disks. This lets you read/write to several hard disks concurrently and speed things up. In other words, that 2GB (or whatever it is) file you are trying to stuff on one disk gets written to 3 or 4 all at the same time.

[HarmsWay]

Hmmm.... that's reminds me. I've been meaning to check my drive for fragmentation after a year of recording with Audacity.

[drumpriest]

Crokett, no need for a raid when recording the # of tracks that I am. Just hammered the crap out of that drive for days on end recording tons of small takes for things. I'm surprised that the fragmentation was that bad, but I wasn't surprised that it was bad, in general.

There's definately something going on with that particular project though, as after a wipe, defrag, and recopy back, and fragmentation down to 2% on the drive, it was still dropping out in that project. No other project has issues at this point, so there must be some setting in Sonar that I'm missing.

btw, overall Sonar producer is a pretty crazy nice tool.

[Raffi]

Yeah,

You gotta keep in mind that video and audio recording is very different from say a Photoshop file or desktop publishing file.

Programs can also play "dirty" with writing files in order to squeeze performance from ordinary discs.

It is recommended procedure to erase these types of drives with some consistency.

Raffi

[drumpriest]

I have no doubt Raffi. I would imagine that the fragmentation is a function of the writting process of the tracks being recorded. You want to avoid seeks during that process, as they are relatively slow, so if Sonar is writting 8 files (of 24 bit 96khz audio), it is probably interleaving the data on the disk in some fashion. That would make sense.

Easy fix is to just copy the project folders to a new drive (which we do for backup anyway), then copy back. This will write the files in a contiguous fashion, eliminating fragmentation much faster than speed disk, for instance.