Sign in with
Sign up | Sign in
Your question

Raid6 Frequent Rebuild, is that normal?

Last response: in Storage
Share
August 14, 2011 3:44:40 PM

Hello, I have a RocketRaid3560 24port sata raid card installed in my Hackintosh system running snow leopard 10.6.5, with 20 internal WD 2TB drive installed as one Raid6 Volume.
The system has been up and running for 8 months now, most of time running smoothly. But 10 days ago, the Raid6 failed suddenly, and report one HardDrive failed, but after restart the system, the failed drive back online and raid card start to rebuild. After about 20 hours, rebuild finished and everything's alright.
Today, the raid alarm again, the same drive which failed 10 days ago, fails again. But after restart, the drive back online as normal and raid6 start to rebuild. As far as by now, the rebuilding work seems fine.

So, what's really happened to my system? Do I need to change that failed disk? In the raid card management software, I read the s.m.a.r.t parameters, seems everything's fine.

Please help me on this issue.
a c 82 G Storage
August 14, 2011 4:23:09 PM

Quote:
Do I need to change that failed disk?
You definitely should replace it. I presume that you're using RE4 drives?
m
0
l
August 14, 2011 4:27:32 PM

I'm using WD AV-GP (the Green Drive).

So, that's means using a hard drive in Raid system, will likely to fail much quicker than using as normal single drive?
m
0
l
Related resources
Can't find your answer ? Ask !
a c 82 G Storage
August 14, 2011 5:21:44 PM

You can expect a higher failure rate (drive being dropped of the array) when using non-RAID hard disks in a RAID because they usually don't support TLER. I presume that you already read http://wdc.custhelp.com/app/answers/detail/search/1/a_i.... It doesn't mean that the AV-GP won't work (it's a good drive though using variable speed drives in a 20 disk RAID isn't ideal), but they aren't supported in RAID other than a 2 drive RAID 0 or RAID 1. Since you implemented RAID 6, just make sure that you replace drives that drop out of the array instead of setting them back online and you should be safe.
m
0
l
August 15, 2011 12:46:11 AM

Thank u for ur detailed information and knowledge about raid drive. So here's the Folowing problem, how can I identify which one is the failed drive? There's ID disk function in raid card menu, it says there would be LED flashing, but can't find it anywhere.
m
0
l
a c 82 G Storage
August 15, 2011 1:03:40 AM

You get LED flashing only if the drives are installed in a compatible RAID enclosure that includes LEDS for each hard disk. When you select the ID disk function in the controller's menu, you should see a LED flashing identifying the selected hard disk, etc. In addition, you should know which drive failed as the controller identifies them by channel. The manual should clearly explain how hard disks are connected and identified. If the documentation is unclear, then contact the manufacturer because you need to know that info, i.e., you can't remove the wrong hard disk if one has already dropped out of the array.
m
0
l
a b G Storage
August 15, 2011 6:49:48 AM

I would gradually replace your drives with something enterprise class. While RAID 6 offers better redundacy, using drives not made for RAID is going drastically increase your failure rate. RAID 6 gives you a little bit of time to recover from a failed drive since your have double parity, but I'd keep a hotspare in there now and replace the drives. All it takes is the unlucky event of a 2nd failure before you can replace the first drive and your whole array is toast. TLER is extremely important in larger arrays since the lieklyhood of an individual sector problem goes up with the number of sectors in the array.
m
0
l
August 15, 2011 7:54:24 AM

GhislainG said:
You get LED flashing only if the drives are installed in a compatible RAID enclosure that includes LEDS for each hard disk. When you select the ID disk function in the controller's menu, you should see a LED flashing identifying the selected hard disk, etc. In addition, you should know which drive failed as the controller identifies them by channel. The manual should clearly explain how hard disks are connected and identified. If the documentation is unclear, then contact the manufacturer because you need to know that info, i.e., you can't remove the wrong hard disk if one has already dropped out of the array.


I will contact the manufacturer, and hope they can clarify the problem for me.
m
0
l
August 15, 2011 8:06:47 AM

tokencode said:
I would gradually replace your drives with something enterprise class. While RAID 6 offers better redundacy, using drives not made for RAID is going drastically increase your failure rate. RAID 6 gives you a little bit of time to recover from a failed drive since your have double parity, but I'd keep a hotspare in there now and replace the drives. All it takes is the unlucky event of a 2nd failure before you can replace the first drive and your whole array is toast. TLER is extremely important in larger arrays since the lieklyhood of an individual sector problem goes up with the number of sectors in the array.


Tokencode, what's your recommendation for replacing the disks? What about the WD RE4?
m
0
l
a b G Storage
August 16, 2011 7:19:39 PM

log-in the RAID GUI... check the S.M.A.R.T. parameters It will tell you what is going on with the HDD.
I would replace the HDD and would not bother with RE drives
Keep in mind the HDD failure rate is up, due to density of the data. I would have couple spare just in case...

As the money concern the price is 2 to 1, so two Green Drive you would have 6yrs warranty in stead of 5yrs :-)
Another way of look at thing
m
0
l
!