SMART status on 3Ware 7500-8?, old drives when to replace?

[MaxBurn]

Would love to replace these with either 4 or 6 120Gb drives, don't want to spend the money though. Please treat the pole like you were spending your own money and all you wanted was reliable data storage. Speed isn't really a factor in this case.


I have eight 60Gig WD drives in my server in RAID5. Four of those drives are getting really long in the tooth and it occurred to me that checking SMART values would be a good idea to see if they are changing. Is this even possible, all the info that the 3Ware monitor utility gives is OK.

I have verify and media scan scheduled for about an hour every week, no problems found ever.

SCSI ID 2 3ware 7500-8 ATA RAID Controller
Array Unit 0 Striped with Parity 64K (RAID 5) 420.14 GB OK
Port 0 WDC WD600LB-00DNA0 60.02 GB OK
Port 1 WDC WD600LB-00DNA0 60.02 GB OK
Port 2 WDC WD600BB-00CAA1 60.02 GB OK
Port 3 WDC WD600JB-00CRA1 60.02 GB OK
Port 4 WDC WD600AB-00BVA0 60.02 GB OK
Port 5 WDC WD600AB-60BVA0 60.02 GB OK
Port 6 WDC WD600AB-00BVA0 60.02 GB OK
Port 7 WDC WD600AB-60BVA0 60.02 GB OK

Two of the drives here are getting close to the 5 year mark and another two are likely close to 4 years (if I remember right). The other 4 are about 8 months old. The old drives are 5400rpm units and the new ones are 7200rpm.

At what point should I think about replacing these drives? They have been running constantly all this time but don't get accessed much, and when they do it's not to heavy on the seek.

 

[Buck]

Why change anything? If the drives work, continue to use them. Isn't that the purpose of RAID 5 - you replace a drive when it fails without the cost of losing your data. Now, if you some how think that it is strategically smarter to replace the aging drives based solely on their age, you can always argue that those 5-year old products have at least proven their reliability. That, of course, is my opinion.

Now, if it were Mercutio, he would replace them all simply because they are Western Digital drives - actually they would theoretically have all failed on him by now, since WD only sends him bad drives.

 

[MaxBurn]

To bad I can't edit polls or posts. I was going to add in the option for don't change anything, wait until something fails.

I think it's funny when people slam on one particular manufacturer, they all mess up here and there. For me WD drives have been great, and between my desktop and this server I have twelve hard drives from WD ranging from those 5400rpm 5 year old jobs all the way up to my new 10krpm 3 month old Raptor 74gig. The biggest drives I own is the WD1000JB's though.

This server in it's first iteration had to have one drive RMA'd right away, it died within a month if I remember correctly, otherwise I don't recall ever having any problems with these drives. That RMA return drive is still in use too. That failure wasn't even a problem as I was running that drive as half of a mirror. It was beautiful, shut it down and took it out RMA'd it and continued to run on the single drive, got it back and rebuilt the mirror and was like it never happened. I think that failure is what sold me on redundant storage in the first place.

 

[MaxBurn]

Oh shoot, and the real question is how do I view the SMART data on these drives? Not possible?

 

[Fushigi]

I also voted to not change anything. If your array is satisfactory

Are you doing a single 8-drive RAID5 with 420GB available or 2 4-disk sets with a total of 360 available? How full is it?

I ask as you're burning more power & generating more noise & heat running 8 smaller drives vs. fewer-but-larger drives. 4 modern 160GB disks in RAID5 would be qieter, faster, produce less heat, and would up your capacity to 480GB after RAID overhead. 200GB or 250GB disks could be used for even more capacity.

 

[blakerwry]

i would probably leave as-is too. But I would have a spare disk on hand incase of failure (and esp multiple failure within a short time)

 

[P5-133XL]

If there are no issues and they do what you need then why mess with them: You can only do harm.

 

[MaxBurn]

Are you doing a single 8-drive RAID5 with 420GB available or 2 4-disk sets with a total of 360 available? How full is it?

It's a single array 420Gig, 391Gig after format and currently 100Gig free.