Image Map Image Map
Page 2 of 4 FirstFirst 1234 LastLast
Results 11 to 20 of 33

Thread: Question about how Intel RAID works

  1. #11
    Join Date
    Jan 2011
    Location
    Vancouver, BC
    Posts
    4,988
    Blog Entries
    3

    Default

    Quote Originally Posted by lutiana View Post
    You are describing an URE (uncorrectable read error), so my guess is you are using drives not suited for RAID use, such as a WD Blue or Green. These drives do not hand off recovery options to the RAID controller and try to do it themselves, and in most cases the drive takes too long to respond to the controllers requests and it then marked as bad and removed from the array. Usually pulling it, and re-inserting it fixes this at least long enough to get the data off. If these are 2TB+ drives you maybe SOL from a data recovery standpoint as the likelihood of another URE while the data is being rebuilt or copies is VERY high, mathematically is approaches nearly 100% with larger drives, this is why RAID5 is not used anymore in the IT field.

    Running data recovery on the drive was a bad idea, but assuming your data recovery did not damage the data on the drive, then just put it back in it's slot and see if the machine can add it back into the array, if so then copy your stuff off ASAP. If the backup has in any way changed the data on the drive, then you're probably toast and should look to your backups ((remember RAID is not backup, it's a high availability technology).)
    They are 1tb intel dc3100s... would have thought those would handle raid ok. I have pulled a few files off successfully but yeah... i thunk the big stuff is probably too compromised. I see companies advertising RAID recovery.. what do they typically do in situations like these?
    Yes, that is how a RAID 0 works, the data is split evenly between both drives, half the file on disk 1, half on disk 2. RAID 10 then makes a duplicate of that on a second set of striped drives.

  2. #12
    Join Date
    Jan 2011
    Location
    Vancouver, BC
    Posts
    4,988
    Blog Entries
    3

    Default

    Quote Originally Posted by dabone View Post
    Yeah, that's a different beast, just software actually.

    The term you need for that is IRST. (Intel Rapid Storage Technology) and there are many different versions of this.

    I've used a utility called raid reconstructor before, but apparently it only seems to handle 0 and 5.

    Diskinternals has a raid 10 recovery software.
    So with a 'software' RAID like Intel's, it needs the OS/driver to actually change anything (sync, resync, write, delete) on the disks themselves, right? I was wondering how you could 'delete' an array in the Intel 'bios' and recreate it without it scrambling everything, but the video and other reading suggests nothing actually happens if the disks are not initialized. I'm just wondering if the controller actually does anything to them independent of the OS/driver. I think from past experience I've gotten messages to the effect that 'rebuilding will occur within OS' actually from the controller.

    I'm willing to invest a bit more effort on this - nothing to lose really but maybe learn something. Just don't want to waste time if there's a likelihood of the drives having had new stuff written to them, scrambling the data. Far as I can tell, the two software I tried (stellar recovery and testdisk) don't actually do anything destructive, unless you've reinstalled your OS and the program on top of the drives (ie you reformmatted them). In my case, I've simply loaded the OS onto a separate drive, and have not given the command to init the disks, just allowed the software to discover the raid and try to read from it.

  3. #13
    Join Date
    Sep 2008
    Location
    SE MI
    Posts
    4,791
    Blog Entries
    6

    Default

    Quote Originally Posted by falter View Post
    So with a 'software' RAID like Intel's, it needs the OS/driver to actually change anything (sync, resync, write, delete) on the disks themselves, right? I was wondering how you could 'delete' an array in the Intel 'bios' and recreate it without it scrambling everything, but the video and other reading suggests nothing actually happens if the disks are not initialized. I'm just wondering if the controller actually does anything to them independent of the OS/driver. I think from past experience I've gotten messages to the effect that 'rebuilding will occur within OS' actually from the controller.

    I'm willing to invest a bit more effort on this - nothing to lose really but maybe learn something. Just don't want to waste time if there's a likelihood of the drives having had new stuff written to them, scrambling the data. Far as I can tell, the two software I tried (stellar recovery and testdisk) don't actually do anything destructive, unless you've reinstalled your OS and the program on top of the drives (ie you reformmatted them). In my case, I've simply loaded the OS onto a separate drive, and have not given the command to init the disks, just allowed the software to discover the raid and try to read from it.
    Other than the fact that "I just wanna use a RAID", I've found them to be a PITA. I used to have a Raid 0 setup for gaming because it offered some gain in speed and efficiency. Now that TB drives are somewhat inexpensive, I find it much easier just to do a simple backup and/or make an ISO of the system. I had a disk failure on one side of the RAID 0 and it took quite a while to rebuild. Yes, I know there are redundancies, mirrors, and all of that, but none of it beat a complete backup. They are fun to play with however.
    Surely not everyone was Kung-fu fighting

  4. #14
    Join Date
    Mar 2009
    Location
    Dublin, CA USA
    Posts
    2,923
    Blog Entries
    1

    Default

    Quote Originally Posted by Agent Orange View Post
    Other than the fact that "I just wanna use a RAID", I've found them to be a PITA. I used to have a Raid 0 setup for gaming because it offered some gain in speed and efficiency. Now that TB drives are somewhat inexpensive, I find it much easier just to do a simple backup and/or make an ISO of the system. I had a disk failure on one side of the RAID 0 and it took quite a while to rebuild. Yes, I know there are redundancies, mirrors, and all of that, but none of it beat a complete backup. They are fun to play with however.
    You can't rebuild a RAID 0 if a drive fails, there is no redundancy in that RAID level. RAID 10 solves that by giving you the redundancy of RAID 1 and the speed of RAID 0. But you also seem to imply that backups = RAID in some way, they don't. RAID is a high availability technology, the data continues to be available if there is a drive failure, only really useful if you are in a position where downtime will cost production time and therefore money. Point is, you'd need both in most environments, save maybe a home environment, where downtime is not an issue, and in that situation you are right and all you really need is a solid and reliable backup.

    IBM 5160 - 360k, 1.44Mb Floppies, NEC V20, 8087-3, 45MB MFM Hard Drive, Vega 7 Graphics, IBM 5154 Monitor running MS-DOS 5.00
    IBM PCJr Model 48360 640kb RAM, NEC V20,, jrIDE Side Cart, 360kb Floppy drives running MS-DOS 5.00
    Evergreen Am5x86-133 64Mb Ram, 8gb HDD, SB16 in a modified ATX case running IBM PC-DOS 7.10

  5. #15
    Join Date
    Mar 2009
    Location
    Dublin, CA USA
    Posts
    2,923
    Blog Entries
    1

    Default

    Quote Originally Posted by falter View Post
    They are 1tb intel dc3100s... would have thought those would handle raid ok. I have pulled a few files off successfully but yeah... i thunk the big stuff is probably too compromised. I see companies advertising RAID recovery.. what do they typically do in situations like these?
    You've had the worst two drives fail, and I am not too sure you'd be able to recovery them unless you can get that drive that did not fail but fell off the array to be recognized and adopted back into the array. I'd guess recovery companies would pull as many bits from the drives as they can, then stitch them back together as best they can using a highly specialized skill set and tool set.

    If I were you, I'd just walk away from it as it is unlikely you'd be able to get much data recovered.

    IBM 5160 - 360k, 1.44Mb Floppies, NEC V20, 8087-3, 45MB MFM Hard Drive, Vega 7 Graphics, IBM 5154 Monitor running MS-DOS 5.00
    IBM PCJr Model 48360 640kb RAM, NEC V20,, jrIDE Side Cart, 360kb Floppy drives running MS-DOS 5.00
    Evergreen Am5x86-133 64Mb Ram, 8gb HDD, SB16 in a modified ATX case running IBM PC-DOS 7.10

  6. #16
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    2,535

    Default

    Quote Originally Posted by lutiana View Post
    But you also seem to imply that backups = RAID in some way, they don't. RAID is a high availability technology, the data continues to be available if there is a drive failure, only really useful if you are in a position where downtime will cost production time and therefore money.
    Over the history of the Internet there has been more than one startup company that's collapsed after massive customer data loss incidents because they confused RAID with backup. If your only copy of the data is the "live" one it just takes one errant software command to wipe it all clean no matter how many drives it's spread and mirrored across.
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  7. #17
    Join Date
    Sep 2008
    Location
    SE MI
    Posts
    4,791
    Blog Entries
    6

    Default

    Quote Originally Posted by lutiana View Post
    You can't rebuild a RAID 0 if a drive fails, there is no redundancy in that RAID level. RAID 10 solves that by giving you the redundancy of RAID 1 and the speed of RAID 0. But you also seem to imply that backups = RAID in some way, they don't. RAID is a high availability technology, the data continues to be available if there is a drive failure, only really useful if you are in a position where downtime will cost production time and therefore money. Point is, you'd need both in most environments, save maybe a home environment, where downtime is not an issue, and in that situation you are right and all you really need is a solid and reliable backup.
    Excuse me, but why can't I REBUILD my "RAID 0"? One just starts over from square #1, or am I only allowed to have one shot and if it gets borked I'm just plain out of luck? I never said a RAID equals anything. What I said or implied was, in my opinion, backups are an easier way to backup and/or store data. Apparently some feel the same way:

    https://www.pugetsystems.com/labs/ar...rible-Idea-29/

    Also Mr. Super Moderator, why are you pushing my buttons with snarky comments? I'm entitled to my opinions just like anyone else on this forum.
    Surely not everyone was Kung-fu fighting

  8. #18
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    2,535

    Default

    Quote Originally Posted by Agent Orange View Post
    Excuse me, but why can't I REBUILD my "RAID 0"?
    When you're talking about RAIDs "Rebuild" has a specific meaning, which is "reconstruct the data on a failed drive from the parity or mirror information on other members of the raid set". Since a "RAID 0" is a striped configuration where half of the total information is stored on each drive and there *is* no parity or other mechanism available to preserve data in the event of a drive failure, IE:

    RAID 0:

    RAID 0 consists of striping, but no mirroring or parity. Compared to a spanned volume, the capacity of a RAID 0 volume is the same; it is the sum of the capacities of the disks in the set. But because striping distributes the contents of each file among all disks in the set, the failure of any disk causes all files, the entire RAID 0 volume, to be lost. A broken spanned volume at least preserves the files on the unfailing disks. The benefit of RAID 0 is that the throughput of read and write operations to any file is multiplied by the number of disks because, unlike spanned volumes, reads and writes are done concurrently; the cost is complete vulnerability to drive failures. Indeed, the average failure rate is worse than that of an equivalent single non-RAID drive.
    So, yes, you cannot "Rebuild" a RAID 0 unless you meant that you replaced the bad drive and restored the contents from a backup. If that's not what happened, well, are you certain you weren't actually running a RAID 1?

    RAID 1:

    RAID 1 consists of data mirroring, without parity or striping. Data is written identically to two or more drives, thereby producing a "mirrored set" of drives. Thus, any read request can be serviced by any drive in the set. If a request is broadcast to every drive in the set, it can be serviced by the drive that accesses the data first (depending on its seek time and rotational latency), improving performance. Sustained read throughput, if the controller or software is optimized for it, approaches the sum of throughputs of every drive in the set, just as for RAID 0. Actual read throughput of most RAID 1 implementations is slower than the fastest drive. Write throughput is always slower because every drive must be updated, and the slowest drive limits the write performance. The array continues to operate as long as at least one drive is functioning.
    That is why he said you "can't rebuild a RAID 0": the "R", which usually stands for "redundant", simply does not apply to RAID 0s. There's nothing to rebuild after a drive failure, half the data is lost, full stop.
    Last edited by Eudimorphodon; July 13th, 2020 at 02:06 PM.
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  9. #19
    Join Date
    Sep 2008
    Location
    SE MI
    Posts
    4,791
    Blog Entries
    6

    Default

    Quote Originally Posted by Eudimorphodon View Post
    When you're talking about RAIDs "Rebuild" has a specific meaning, which is "reconstruct the data on a failed drive from the parity or mirror information on other members of the raid set". Since a "RAID 0" is a striped configuration where half of the total information is stored on each drive and there *is* no parity or other mechanism available to preserve data in the event of a drive failure, IE:

    So, yes, you cannot "Rebuild" a RAID 0 unless you meant that you replaced the bad drive and restored the contents from a backup. If that's not what happened, well, are you certain you weren't actually running a RAID 1?

    That is why he said you "can't rebuild a RAID 0": the "R", which usually stands for "redundant", simply does not not apply to RAID 0s. There's nothing to rebuild after a drive failure, half the data is lost, full stop.
    Well, maybe the semantics in your IT world would call for such. But here in my shop (upstairs 3rd bedroom) rebuild is rebuild. I replaced a failed HD and "rebuilt" my RAID 0. Have a look:

    http://www.freeraidrecovery.com/libr...d-rebuild.aspx

    Possibly referring to "RAID recovery"?
    Surely not everyone was Kung-fu fighting

  10. #20
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    2,535

    Default

    Quote Originally Posted by Agent Orange View Post
    Well, maybe the semantics in your IT world would call for such. But here in my shop (upstairs 3rd bedroom) rebuild is rebuild. I replaced a failed HD and "rebuilt" my RAID 0. Have a look:

    http://www.freeraidrecovery.com/libr...d-rebuild.aspx
    From the "RAID Recovery Guide" linked on that page:

    RAID 0 Recovery

    Before you rush to recover RAID 0, you need to understand the causes of RAID 0 failure. There are two different types of RAID 0 failure:

    failure of one or several RAID 0 member disks;
    failure not associated with the member disks.

    RAID 0 member disk failure

    Since RAID 0 arrays are non-redundant, then if one of the member disks fails, then data that was on the failed disk is lost forever. Having data from the rest of the member disks you can try to recover files. However, only the files which are smaller than (N-1)*(block size) can be recovered. Even files smaller than that limit can be unrecoverable if the part of the file happens to be on a failed disk.

    So in general, if one of the member disks fails beyond repair, it is impossible to recover data from RAID 0.
    IE, it says exactly what I said, that if a disk fails to the point of being unreadable then you cannot "rebuild" a RAID 0. Full stop.

    Now, at this point I'm going to grant that a "failed disk" isn't a failed disk. Depending on the RAID controller sometimes a disk will be marked "dead" and the RAID taken down the moment a disk starts throwing enough S.M.A.R.T. errors. In these circumstances you might be able to use recovery software designed to tolerate/ignore these errors to copy enough of the data from the damaged disk to another one to get things working again with minimal data loss; that's apparently what you did. But this is not a normal "Rebuild" because, again, when you're talking about RAIDs that term solely applies to recovering from drive failure conditions the RAID level you're using is designed to tolerate. What you did was recover your RAID. I realize that informally the semantics don't matter to you, but this *is* a tech forum.

    (* My favorite story about S.M.A.R.T. errors: the PERC RAID controllers in old Pentium III/4 era servers like the 2550 had a fun bug back in 2002-ish; the RAID firmware had a function for logging SMART errors in RAM so they could be read by monitoring software running on the host OS to warn of potentially pending doom, but in normal circumstances the RAID wouldn't actually *fail* a drive and switch to degraded mode unless something truly catastrophic happened. The problem was that the way they wrote the handler for going to degraded mode if any of the other drives had logged a SMART error, even a transient one, since being powered up the controller would while assessing the situation decide those errors were fatal and fail that drive too, even if it resulted in killing the whole volume. I witnessed this cascade failure happen right in front of me, the drives pinging off one at a time as they were declared kaput.)
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •