Link to home
Start Free TrialLog in
Avatar of Soupbone79
Soupbone79

asked on

data drive that 82 percent fragmented with 1.8 terabytes of data

I have a client with a windows 2012 server r2 with data drive that fragmented with 1.8 terabytes of data. It's taking forever to defrag, i mean days. This is very nice server with raid 5 (LSI card with 6gbs support and all shows optimal in the web gui) Two 8 core Zeon CPU's, 64 gigs of ram. it has has 6 1 terabyte seagate serial attached scsi drives in it. it has been running fine for a long time. Now it's not. when I noticed the drive was fragmented at 82 percent. Has anyone got any ideas to share?
Avatar of Cliff Galiher
Cliff Galiher
Flag of United States of America image

What tool are you using to get that statistic?  When you say it "was" running find and now is not, what are the actual symptoms?  I mean, we can random conjecture, but it really would be random.
Avatar of Member_2_231077
Member_2_231077

You're gambling with RAID 5 on 6 reasonably large disks, if one fails there may be a bad block on one of the others leading to data corruption and also it will take quite a while to rebuild during which time you have no parity at all. You should really be using RAID 6 even if it costs an extra disk. You can migrate on the fly but it will take days.

Check the state of the battery in the LSI MegaRAID Storage Manager. if that is dead it will slow to a crawl with RAID 5 or 6.
Avatar of Soupbone79

ASKER

It does not have a battery plugged into the board could it be somewhere else? I do not see it listed in the intel raid web console
Which controller model no. is it? You said LSI and Intel so far.
G27504-601
Cheap card doesn't have cache, but there again it didn't used to be slow so probably something else.
yes I keep coming backed to fragmented. There no viruses nor malware rootkits ransomeware and anything like that I have confirmed that.
How full is the drive?

For any applications where disk speed is important, you really want to keep 10% free as an absolute minimum. With more than 10% free space, NTFS usually manages not to fragment too much. When drive space is getting low, and users keep having to find old content to delete each week to make room for new files, then fragmentation can quickly become problematic.

Having said that, it takes a LOT of fragmentation to invoke any measurable performance problem, and it depends of course on how fragmented commonly used files are, not average fragmentation.

I suggest you clear at least 20% disk space and run a defrag over a weekend. Once fragmentation is somewhat under control, if you ensure the drive is never more than 90% full, a quick defrag every few weeks should keep fragmentation minimal.

It may be worth grabbing a "trial version" of Diskeeper, If fragmentation is spectacularly bad; this can defrag the swap space and MFT, which Windows cannot, and is sometimes better at tidying up a real mess.
1 TB  drives are still ok in RAID 5.  Rebuild times are passable, assuming you're not using the volumes heavily during rebuild, but you are reaching a point where a second disk can fail soon after the first disk does.  http://www.enterprisestorageguide.com/raid-disk-rebuild-times.

Don't defragment your disk if you don't have free space left.  It will take forever even with just 1% free space.  You need enough free space to allow the deframenter to move all fragmented blocks int contiguous spaces, to make more contiguous space.  If you don't have enough free space, you might as well stop your defrag right now.  If your drives are starting to underperform, possibly due to S.M.A.R.T. failures, you're only going to speed up the eventual failure with all that activity.

The quicker, and somewhat easier way to defrag, is to buy an external 5TB, 6TB, or 8TB disk.  Copy or move off all the data to it, so that you have a sequential copy of all that data, then rebuild or erase the RAID drive and copy it back.  You might as well get the external disk to have a backup of the data anyway, since your 1TB disks are probably quite old.
Hey SerialBand,

The consensus position has always been anything over approx 700GB is not a good idea - admittedly 1TB is less of a risk than 11TB spindles, but its still a risk
It's approximate.
the 700GB is an approximation - 1TB is another 45%
This question needs an answer!
Become an EE member today
7 DAY FREE TRIAL
Members can start a 7-Day Free trial then enjoy unlimited access to the platform.
View membership options
or
Learn why we charge membership fees
We get it - no one likes a content blocker. Take one extra minute and find out why we block content.