Unsolved

This post is more than 5 years old

84 Posts

38615

November 15th, 2006 14:00

How to check RAID1 parity

Hi.

I have 2 500 GB harddisks as RAID1 in my XPS700. Normally, the nVIDIA RAID controller only reports errors when the files are accessed. Can I tell the nVIDIA RAID controller to make a check of the data integrity, i.e. make it compare the complete content of both disks?

Usually, hardware RAID controllers offer a funktion called "parity check", which checks the complete RAID array for errors. I case there's a problem, the faulty disk can be replaced.

Is that also somehow possible with the nVIDIA RAID controller?

Thanks,

2 Intern

 • 

2.6K Posts

 • 

7.9K Points

November 15th, 2006 14:00

Not to laugh but there is no parity creation or checking in Raid 1 or 0.
 

2 Intern

 • 

2.6K Posts

 • 

7.9K Points

November 15th, 2006 15:00

There are two utilities the regular nvidia RAID program and it does not say anything unless a drive has a problem with smart or is dead and ntune which is basically the same.
SInce you have mirroring it operates like you have one drive so you really don't have to check anything, because if one drive dies you will get a notice at boot up for sure and maybe the program will open if you have it in the tray.
Since its a very low level controller there is not a lot of monitoring availalbe.

84 Posts

November 15th, 2006 15:00

Of course, you are right (I spoke of an RAID5 array which I also troubleshoot at this time).

What I meant is ... is tehre any mean to check whether (1.) all data on both disks is readable and (2.) matches (=both disks have same content)?

Sorry for the wrong formulation in the initial posting.

Thanks,

2 Intern

 • 

2.6K Posts

 • 

7.9K Points

November 15th, 2006 16:00

nvidias ntune.
You can run chkdsk or any drive utility you want, your drives are mirrored in real time.
If you want more than that you will have to buy a real SAS/SATA raid controller and spend the few hundred dollars for it.
There is no way to check since there is nothing to check. With real time mirroring you would notice the same issues with a bad drive as if it were a single drive.  There is no comparision for onboard RAID to a several hundred to few thousand RAID controller and you don't need it either.

84 Posts

November 15th, 2006 16:00

I only know MediaShield. What is the second tool?

That's exactly my point. SMART doesn't catch all errors. Hence, RAID hardware manufacturers usually build include some kind of parity check (RAID 4,5,6) or data integrity/comparison check (RAID 1) into their software, which physically verifies the complete array. This takes sometimes many hours, but at the end, array functionality is assured.

I don't HAVE to check anything, but (worst case scenario) both brives could have medium errors (read error) for the same file, hence this would be gone! Any ... it would only be discovered when accessing the erroneous file.

This situation is trapped by performing regular parity/comparisopn checks of the array.

Even if the controller has no monitoring features, is there no was to check integrity (= compare data of both drives)?

Thanks,

84 Posts

November 18th, 2006 03:00

Is ntune harddisk related? As far as I could determine, it doesn't change/check anything for the RAID array.

Since 2 RAID1 drives are considered as one, it ol´nly checks the file system integrety. Reads are done alternatively between the 2 drives. If there is a data discrepancy between the 2 drives, and chkdsk takes the "good" drive to load this chunk of data, it doesn't report any error. Anyway ... it would only be a file system integrety error. chkdsk doesn't compare the data of both drives (to the best of my knowledge).

> There is no way to check since there is nothing to check.

There is something to check. If one harddrive has a medium error (e.g. read error), this one will only be detected after the file, which contains the error, is read. In this case, you can change the drive and rebuild the data from the "good" drive (= rebuild the array RAID1). Imagine now, if that one also has a problem at another sector, then you can't rebuild the array. That's my point.

A data consistency check would possibly spot an error at one disk, before the second disk also creates a problem. That's exactly what happened to one of my RAID5 arrays (4 disks). During the rebuild of drive 2, drive 4 reported a problem!

Regards,

2 Intern

 • 

2.6K Posts

 • 

7.9K Points

November 18th, 2006 05:00

You need to find and read the operating technical manual for it. You are trying to make it have more capability than it has and you don't understand how it works and cannot compare it to a real controller of any raid level except in the most basic sense.
Writes do not alternate as in your thinking and that is just for starters. They occur at near the same time but the timing can be different but you have SATA which is a point to point bus. There is absolutely no difference for writes compared to a single drive with this type of raid except that there is two of them but there is no additional checking other than what the controller would do for one drive.
No Events found!

Top