Sign in with
Sign up | Sign in
Your question

Raid Level 1 Question

Tags:
  • Hard Drives
  • NAS / RAID
  • Boot
  • SATA
  • Storage
Last response: in Storage
Share

Do You build your own pcs?

Total: 10 votes

  • No
  • 0 %
  • Yes, Of course
  • 20 %
  • I build them for my friends/family too
  • 80 %
August 11, 2006 3:57:55 AM

If you have 2 SATA drives in RAID 1 and one fails...can you just boot as SATA instead of raid and everything will work fine?...I understand that RAID 1 is mirroring, but if one fails can you really just change the bios to SATA boot and will it boot without having to do anything else?...thank you

More about : raid level question

August 11, 2006 4:48:33 AM

The drives will still be in RAID1 you don't change that. As long as you set your boot.ini to point to both and make sure the bios is set to "boot to other devices", you won't have a problem.

If you change the bios to sata, the raid metadata won't be read and it might not boot. Some controllers drives will boot like that, not all. And if the RAID is working properly, the system won't go down, you'll just get a notice from the raid software saying a drive dropped, you did install the monitor right? :) 
August 11, 2006 2:50:16 PM

The raid state in any mirror set up will boot ( as long as it is not a controller issue) in whats called a degregated state. The mirror or primary drive will work independently from the other until some type of proactive action is taken. Usually your can rebuild the drive, or replace the drive then rebuild the mirror. You are also limited to the data corruption and timing because most raid setups, like enterprise systems, drives mirror on a staggered basis or delay. So you'll find that there may be data with time stamps that are off or don't match the original.
Related resources
August 11, 2006 3:44:15 PM

Quote:
You are also limited to the data corruption and timing because most raid setups, like enterprise systems, drives mirror on a staggered basis or delay. So you'll find that there may be data with time stamps that are off or don't match the original.


Where have you heard this?

Data that is copied to another drive with a time delay like you're talking about is an online backup system or "near-line" backup system. That's not a RAID.

RAID-1 is not called a mirror for nothing. Both drives maintain an exact copy of each other's data at all times. No delay. The controller performs simultaneous writes to both drives, always. The controller does not report a successful write back to the OS until both drives have completed the write (ignoring write-back cache optimizations, of course).

I have 6 servers at my company, all run RAID-1 for their C: drive. Any of those drives can fail, no problem. Replace the drive, controller will transparently rebuild the mirror to an exact copy of the data on the good drive. Some of the controllers even support hot-swap and hot-rebuild. Don't even have to reboot the server. Ultimate in redundancy.

There is never any sort of issue with "time stamps" or "data corruption" when recovering on a RAID-1.
August 11, 2006 4:10:10 PM

A couple of my servers have dual raid1's and I have hotspare's in a 5th hotswap bay, I'll get an email about a failed drive and I don't even need to worry, the hotspare immediately rebuilds and I can replace the failed drive at my leisure. Course I've only had this happen twice on all of my servers over the years, but it's nice to have it and know it works. Keeping a space scsi drive around isn't exactly cheap though (currently 15krpm 147GBers, one in each of 3 servers at one location alone).
August 11, 2006 5:11:52 PM

Quote:
You are also limited to the data corruption and timing because most raid setups, like enterprise systems, drives mirror on a staggered basis or delay. So you'll find that there may be data with time stamps that are off or don't match the original.


Where have you heard this?

Data that is copied to another drive with a time delay like you're talking about is an online backup system or "near-line" backup system. That's not a RAID.

RAID-1 is not called a mirror for nothing. Both drives maintain an exact copy of each other's data at all times. No delay. The controller performs simultaneous writes to both drives, always. The controller does not report a successful write back to the OS until both drives have completed the write (ignoring write-back cache optimizations, of course).

I have 6 servers at my company, all run RAID-1 for their C: drive. Any of those drives can fail, no problem. Replace the drive, controller will transparently rebuild the mirror to an exact copy of the data on the good drive. Some of the controllers even support hot-swap and hot-rebuild. Don't even have to reboot the server. Ultimate in redundancy.

There is never any sort of issue with "time stamps" or "data corruption" when recovering on a RAID-1.

Raid-1 is mirroring, call what you want, but it is. What I was referring too, was when a system begins to fail and the drive that is currently being read as the primary degrades in functionality there is a helm of data corruption that is written to the second, mirrored drive. When failover or redundancy occurs, you now are working off a drive that has corruption written to it. It happens many times, depending on how the primary drive goes down. I was not implying that this is what happens every time an array is broken.
August 11, 2006 6:25:10 PM

If you have a decent quality RAID controller, it will probably work fine as a single disk in "degraded" mode. However, there is no such guarantee - especially with cheap PCI RAID cards.

My experience with a cheap card and 2 80 gig drives in RAID 1:

While testing, I removed the primary disk from the array using the RAID card's own utility (not a physical disconnection). It should have been a clean and easy disconnect, but almost instantly I got a BSOD in Windows XP Home Edition. The RAID array would not boot again, would not rebuild, and neither of the drives would boot up by themselves. I tried to put each drive in another system, and they both showed up as unpartitioned. The final resolution was to rebuild the system - isn't that what RAID is supposed to help avoid instead of cause?

On the other hand, on my other PC, the RAID 1 array connected to my ASUS MB tolerated pulling the power cord on a drive, the system stayed up, and was able to subsequently reboot.

My advice would be that YMMV depending on your RAID controller, and it would be a good idea to test your RAID controller's capabilities and quirks with non-essential data/drives BEFORE the array crashes on its own and you have to try to recover the data.
August 11, 2006 9:03:39 PM

I've manually failed both my 1640 and 2320 (boy does that 2320 have a loud alarm), they are both graceful, no os issues, and the raid 1 boots off the second drive no prob. Same with some nvraid's I've used, never had an os crash on me.
!