Topic: You Can't Be Serious (Read 4596 times)

benn600 · « **on:** February 11, 2008, 07:35:20 am »

What a surprise! I woke up to a failed hard drive again! As I walked by the server I noticed a light was solid. BAD SIGN. So turns out drive 15 failed in the night. Why do they always fail in the night? Why do they fail so often? This is ridiculous! I replaced the drive with another one I have laying around--which is one that was previously FAILED. I can't be replacing drives every 2 months! That's $50/month on new hard drives! What is going on??? It's rebuilding with the replacement drive.

Mr ChriZ · « **Reply #1 on:** February 11, 2008, 09:29:23 am »

$:-\$ Something can't be right

The last hard drive problem I had was in a 386 DX.
I'm surrounded by the things now...
Maybe I'm off balencing my hard drive luck on you!

robydago · « **Reply #2 on:** February 11, 2008, 10:21:10 am »

i used cheap and "single" drives for years in our dirty cheap servers (actually PCs used as servers) with no problem at all.

3 years ago we upgraded our servers to server level ASUS motherboards with integrated RAID controllers so we started using mirror RAID... and we started having a lot of problems, with some HD failures and many RAID rebuilds for more than 1 year...
I have no idea why, but after that period and without doing anything, things are settled and those mirrors are not having any problems now...

But I think will never use RAID anymore, unless I can afford enterprise level RAID controllers and HDDs

KingSparta · « **Reply #3 on:** February 12, 2008, 01:03:25 pm »

Quote

unless I can afford enterprise level RAID controllers and HDDs

What Does "enterprise" mean?

Does It Come With James T. Kirk?

benn600 · « **Reply #4 on:** February 12, 2008, 06:06:32 pm »

Yea, isn't my card "enterprise"? I suppose there are higher end cards (without a doubt) but it is above personal use!!

bspachman · « **Reply #5 on:** February 12, 2008, 06:29:40 pm »

I don't know--karma? But I can say that I used to go through a spate of bad drives every year until I moved up to the enterprise-class drives. These are a little more expensive, but supposedly are better able to tolerate the 24/7 'always-on' operation found in servers....

No problems at all since moving to them.

For example, in the Western Digital world, see the differences between 'normal':
http://www.westerndigital.com/en/products/Products.asp?DriveID=307

...and 'Enterprise':
http://www.westerndigital.com/en/products/products.asp?driveid=331

Later!
brad

KingSparta · « **Reply #6 on:** February 12, 2008, 07:39:48 pm »

Quote

I don't know--karma?

Not likely

I remember another scheme where the media makers just meant that there maybe less errors on the media\platters Or in other words sectors.

johnnyboy · « **Reply #7 on:** February 12, 2008, 09:11:56 pm »

You pulled all of these out of USB myBook's right?
Try low level formatting them before you put them back in and see if it helps at all.

benn600 · « **Reply #8 on:** February 12, 2008, 11:15:24 pm »

Low level formatting? Oh sure. It's not like I have any important data on the array.

I actually formatted most of them separately although I just used the Windows full format which isn't as good as a low level format.

johnnyboy · « **Reply #9 on:** February 13, 2008, 06:40:42 am »

It's raid 6 - shouldn't be a problem to pull each drive out and do a low level format on it

Just start doing it from now on with each drive that 'fails' and see if it fails again the next time you use it or if that fixes it.

benn600 · « **Reply #10 on:** February 13, 2008, 02:26:40 pm »

True. Wonder if the RAID card could be smart enough to low level format drives as they are added to the array as a replacement drive? Seems like a good idea or option.

KingSparta · « **Reply #11 on:** February 13, 2008, 06:23:50 pm »

Use A Timex Sinclair ZX-80 Computer To Format It

fitbrit · « **Reply #12 on:** February 13, 2008, 07:08:37 pm »

Are you even bothering to send the drives in for warranty replacement/repairs?

How is the cooling in your server(s)?

I hope your luck improves soon.

benn600 · « **Reply #13 on:** February 13, 2008, 11:22:50 pm »

Send them in? They are fine! WD would get mad at me and charge me for sending them perfectly fine drives.

The cooling is incredible. 5 +high+ powered fans. It is a full, legitimate server case. See pictures in another thread.

fitbrit · « **Reply #14 on:** February 14, 2008, 12:04:35 am »

Quote from: benn600 on February 13, 2008, 11:22:50 pm

Send them in? They are fine! WD would get mad at me and charge me for sending them perfectly fine drives.

The cooling is incredible. 5 +high+ powered fans. It is a full, legitimate server case. See pictures in another thread.

Perhaps I misunderstood what a "failed" drive means, then.

benn600 · « **Reply #15 on:** February 14, 2008, 12:26:20 am »

Yea, the problem is that the card fails drives when it doesn't seem like they need to be failed. It appears that they then work fine.

newsposter · « **Reply #16 on:** February 14, 2008, 12:45:06 am »

You need to find out what parameters your card uses to fail out drives. On top of drive failures, it's totally possible that a cabling or power supply issue is causing drives to fail out.

And get a secondary program that will monitor the SMART parameters of your drives.

benn600 · « **Reply #17 on:** February 14, 2008, 11:11:51 am »

I've attempted to ask Promise what parameters are used. I don't recall getting a useful answer. I might not understand it anyway, though.

The error it gives is that there was a timeout and then an address.

ThoBar · « **Reply #18 on:** February 16, 2008, 07:23:54 pm »

The is new new class of drives from WD (I think Seagate have some too) that are designed for RAID use. They may address the exact problems you're experiencing...

(From the link bspachman provided...)

Quote

RAID-specific, time-limited error recovery (TLER) - A feature pioneered by WD, significantly reduces drive fallout caused by the extended hard drive error-recovery processes common to desktop drives.

I would hazard a guess that these drives will almost certainly address the situation you've got. Perhaps see if there is a drive firmware upgrade available for your current drives that would let you adjust the drive recovery time.

I have seen situations similar to your experience also, when using a budget 2 drive mirror card and a pair of desktop drives.

benn600 · « **Reply #19 on:** February 16, 2008, 10:39:22 pm »

So the RAID card requests some data to be written (or read) and the drive takes just a little bit too long, so it thinks it's failing and disconnects it?

I have a suspicion. I always see a failure from 3-7 AM or so. That is right around when my Podcast downloader downloads podcasts. A few of them are bit torrent downloads. I'm very suspicious that the heavy, scattered data writing could be causing the problem. I currently have the files downloaded to a temp directory and then moved via a batch file to a "pretty" location. Juice does not allow custom folders and the default for most podcasts are really long and annoying. So if I move the starting download location to one of the other 3 drives in the system, then it could reduce this error.

I always wondered why I've never seen a failure in the daytime when I'm up and using it! So my theory seems somewhat possible.

KingSparta · « **Reply #20 on:** February 17, 2008, 12:24:33 pm »

Is there something Better Than Raid 6?

Is There A Way To Have Two(2) Raid Controller Cards Save The Data To Two Different Drive Arrays?

Maybe Also Something Like:

http://cooldrives.com/insaiira1mic.html

INTERACT FORUM

Author Topic: You Can't Be Serious (Read 4596 times)