What Happens When Hard Disk Fails in RAID 5

last updated in Categories , , , , , , , , , ,

Q. What happens when a hard disk fails in RAID 5? How many hard drive can fail in RAID 5?

A. RAID is acronym for Redundant Array of Independent Disks, also known as Redundant Array of Inexpensive Disks. It is use to increase reliability. You combine multiple physical hard disks into groups (arrays) that work as a single logical disk i.e. this array of drives appears to the computer as a single logical storage unit or drive.

RAID-5 features

RAID level 5 has following features:

  1. Striping with distributed parity
  2. Fault tolerance supported
  3. Redundancy supported (parity)
  4. Hot spare option supported
  5. Minimum 3 or more disk needed
  6. Performance: Good
  7. Protection: Good (if any disk fails data can be recovered)
  8. Above characteristics makes RAID 5 popular and is well suited to transaction oriented environments.

What happens when hard disk fails in raid 5

  • Because of parity, information all data are available in case one of the disks fails. If extra (spare) disks are available, then reconstruction will begin immediately after the device failure.
  • However if two hard disks fail at same time, all data are LOST.
  • In short RAID 5 can survive one disk failure, but not two or more.

Update: See why RAID 10 is better choice for database server instead of RAID 5 for performance and reliability.

Posted by: Vivek Gite

The author is the creator of nixCraft and a seasoned sysadmin, DevOps engineer, and a trainer for the Linux operating system/Unix shell scripting. Get the latest tutorials on SysAdmin, Linux/Unix and open source topics via RSS/XML feed or weekly email newsletter.

63 comment

  1. i ve 6 disks with raid 5 … Now one of them fails and can i change it directly? Should i have to do anything before replacing the red blinking failed disk ? My server is compaq proliant 1600 ..

  2. If it is hot swappable disk, just replace it. Otherwise shutdown box and replace hard disk. Reconstruction will begin immediately after replacement!

  3. Hi,

    I had raid 5 with three harddrive of 250 gb… out of three one hard drive fail so according to the raid 5 setup we change the fail harddisk and inserted brand new of 500gb and keep the process for rebuild, it took about 8 hours to rebuild but when the procees was completed i was not able to see the drive in the computer so can you please help to let me know of i am missing some configuration or settings…


  4. > Nikhil “harddrive of 250gb” “harddisk new 500gb”
    Mustn’t it be a pair of harddisks with the same size?

  5. In short RAID 5 can survive one disk failure, but not two or more.

    Regarding this statement,

    it is more better to use external storage or mapped network HDD rather than RAID5 arrangement?

    I am a little confused too

  6. If you can find zero erros associated with disk you can bring the hard drives(both) force online .. its your luck if you dont loose any system files during force online and system boots up without any missing system file error(s).

  7. Dear Sir,
    I need the some technical query what is the RED O , RED one and RED 10 please give me

  8. yeah dude ur crazy. it not only needs to be the same make, model, disk size almost everything. i dont even think it put data on that 500, that doesnt sound right

    1. You can replace a failed RAID member with a larger drive, but not a smaller drive. The RAID will only use the space equivalent to the smallest drive in the array. You can also usually replace a failed RAID member with a drive that runs at a different speed than the other array members.

  9. Hi,
    I have built one RAID 5 system using Intel on-board RAID Controller using 3 HDD’s each 1 TB.

    Unfortunately some of Electrical Fluctuations One HDD got Faild (i think that is OS installed)

    Kindly help me in procedure of Replacing the New HDD. Directly can I replace the HDD or before installing HDD I have to follow any procedure to recover the data?

    Kindly help me out in this issue.

  10. Hi,
    We have MS-Exchange2000 having qty.5,18.4GB HDD and 36GB internal HDD. we are running out of HDD space and just left 1GB. also have RAID5. May I replace 18.4GB to 72GB Internal HDD?and can it be detect or configure automatically?

    1. I recommend only one method for this:
      *Notify users of downtime, estimate 8 hours or so, depending on your server speed for backing up/restoring data
      *Take full “image” backup of server to an external media (I use Acronis, but it is spendy), so that it can be restored using a bootable disk
      *Restart the server, boot into the RAID manager and delete the existing RAID array.
      *Shutdown the server
      *Remove old hard drives, be sure to mark your old drives (0,1,2,3,4..) as you pull them out incase you need to revert back to them
      *Install new drives
      *Start server, boot into RAID manager (usually CTRL + M at BIOS) and create the new RAID array for the new drives.
      *Restart and boot to the bootable disk for the imaging application and restore the data.
      *Test that everything is functional and then pat yourself on the back if all is well.
      Don’t forget to buy a hot spare drive that is the same size, speed etc.. and always monitor conditions for heat, humidity, dust… stuff that kills components.

      1. This is the best option if you are running short of space. We did it in our enviroment and it was a great success,

  11. To respond to a couple inquiries, you MUST replace a single drive from a RAID array with the EXACT same drive. So, each drive independently must be the same size, same speed and by the same manufacturer. It must be the SAME.

    You cannot swap one disk that is smaller to one that is newer.

    The only way to upgrade the size of an array without completely rebuilding it is to ADD another drive to the array, thus making it larger across the system, however, when you add another 18.4 GB drive, you will NOT get exactly 18.4GB of additional storage, but a smaller amount based on the RAID array.

    Also, if you wish to move the data over to another array, the old array must stay in tact EXACTLY how it was originally built. In other words, you need to make sure that if you pull a drive out of a certain tray, you need to number and mark it to make sure it goes right back into the same tray. You cannot swap drives around in the array.

    Lastly, if you are going to upgrade an array, my suggestion is to backup your entire system to an external drive by using Acronis or Microsoft’s built in backup system in Server 2008 and once you have a complete backup image, take the RAID down, rebuild it from scratch and then import it from the external drive.

    This is a really easy thing to do if you are running VMware and can simply backup an image of the entire system and drop it onto the external drive and then reinstall VMware and import the image back into the VMware system when you build a new array.

    Just make sure that if you are upgrading the size of the array by building a new one, that you get the exact same drives for the array. I would consider a RAID 10 if you can and if not a RAID 5.

    That’s just my thoughts. You can take them or leave them, but I have had to rebuild arrays a number of times both successfully and not! ;)


  12. hello
    i have 3 hard drives that connected by raid 5 technology one of them had break down , i didn’t find the same thing i found 73GB 15000kbs
    and the old one was 73Gb 10000kbs

    now when i buy the new one and connect it with the the other 2 hard drives ? is it OK ?or i will face conflicts ???

  13. i love this forum and i wil love to say that am new to this RAID stuff and i will appreciate if someone could help me out with full details on it and what to buy in other to set it up for a server. thnks

    1. The server must support RAID technology. There are two methods; hardware or software. Hardware RAID is faster and more fail-safe.
      For hardware RAID, you can boot into the RAID manager at startup and create virtual disks or manage existing disks. Software version is done inside the OS. Most server’s like Dell have an application that will allow you to manage the hardware RAID from within the OS so you don’t have to shutdown the server to make changes. This is called Dell OpenManage Server Administrator.
      To setup a new RAID, you first need to backup all data to an image file (I use Acronis, but it is spendy) that can be restored using a boot disk. Next, you will create the RAID array, which wipes all data from the disks and prepares them for the array.
      RAID 10 is best, as it benefits from speed and redundancy. RAID 5 is second best, it at least has redundancy so you can lose a drive and still be operational. Then you restore the backup image to your server (using boot disk from Acronis). Make sure you have a hot spare on hand for incase a drive fails, and always monitor conditions to be sure server is not too hot, dusty etc.. which perpetuates hard drive failure.

  14. i want some clarification from raid 5 if I have 8 hard drives that connected by raid 5 technology two of them had break down at same time , then raid 5 is work or not ?

  15. I have a Lacie Drive External USB HDD. Chassy contains 4 x 500Gb drives which was operating as Raid 0, one drive has failed on this. Could someone recommend any software that could rebuild the data ? I understand it probably would not be possible to get all data back, but 3/4 of it would be quite nice :D
    Also if there is no software that can recover from a USB Based drive, I’d consider throwing a raid controller into a desktop.
    Also would i need to know which order of drives are inside the Lacie box, and likewise order them in the same logical order in the desktop controller, or does that matter at all ?
    Hopefully this made sense to someone :D lol

    1. Happy to find such an informative / educational site with so much inteligent input into varous scenereos and problems.

      Keep it up. Going to sign up for the weekly newsletter.


  16. the drives does not need to the be the same size. it is required that the new drive to be BIGGER. As for speed., that i have to check but most likely they have to be the same speed.
    As for manufacturer, you do not need that same manufacturer but it is strongly recommended.

    I have personally tested with different drive sizes and manufacturer and have had no problems rebuilding the raid 5. You should however remember to back up.

    Remember, raid means Redundant Array of Independent Disks, In english it means, if one disk goes out your server continues going, it DOES not mean the files are BACKED UP. Remember to backup you files into a seperate mediem.

  17. My RAID 5 config failed somehow. I got on the phone with Dell and they had me try quite a few things with the BIOS and Perci settings. It said all my hard drives were online at one point, then it found 2 foreign, then rebuilt so all were online again. I don’t car about my os, I just need some small data file that was not backed up. 2 computer guys looked at it and said I had little chance of recovering my data, but the file recovery guys say it may be recoverable, everyone keeps tinkering with it and it keeps rebuilding and I know this is damaging. I downloaded “File Scavanger” software and it did recover some data that I needed, but not everything as single drive search. The software searches Raid 5, but the order is jumbled. The Dell Power Edge 2900 is about 3 years old with 4 250G Hard drives. Does anyone think I may be able to recover my data?

    Thanks for your thoughts.


  18. In our office we have two diff DC (domain controlllers) in diff locations with diff Oraganstion Units, I want to import 1 OU with user login account from one DC to another DC please help me.

  19. @ “if 2 drives fail you are hosed … ”

    I am no expert … far from it: DIY master. So, when I saw that my Logical Drive failed this morning, then saw from WebBIOS that 2 of 3 Physical Drives had failed, then read this forum … I was NERVOUS. Then I tried something: in WebBIOS I put the FAILED Physical Drives back ONLINE, then rebooted. At startup I got a blue-screen stop error “INACCESSIBLE_BOOT_DEVICE” with a note to CHKDSK /F. At second reboot, CHKDSK ran automatically. Then ….

    VOILA! I was welcomed by my login screen. Eureka!

    Perhaps I was just plan lucky, but … if you have multiple failed Physical Drives and ARE NOT WORRIED ABOUT LOST DATA SINCE YOU HAVE REGULAR BACKUPS (you really really really should …) then give this workaround a try.

    Note: On a DELL Precision WIN2000 w/ Service Pack 3 “server” (really a workstation…) with a laughable 523,000 K of RAM. HA! Will I be upgrading soon??? You betcha!!!

  20. Raid 5 only can sustain a single lost disk, it doesn’t matter how many disk you have.
    if you have 7 disks with raid 5, the data will still be there if you lose 1 but if you lose 2 you are a died duck. It is expensive and more risky to retrieve information for Raid.

    Usually, from experience, when 1 disk is lost, get ready for more over the period.

  21. I had two hdd out of 3 fail and still got data back, just pointing out that your information is not 100% correct. bye

    1. It cannot be recovered the regular way. The raid is about redundancy, which somehow means having another copy of your data for an online restoring. 2 fails out of 3 drives on RAID 5 won’t by no means work. and It cannot be RAID 6 as RAID 6 requires 4 Drives at least to work.

      Unless you have invented new technology, what you are saying is absolutely incorrect, and it is very rude to just under value other people efforts to share information in here. At least I don’t think that is a nice way of writing opinion.


      1. I also agree with Aymen, as in RAID 5, two out of three disks down means there is no disk left where data can be kept. Like disk1 is having a,b and c data and disk 2 is having d, e and f data. The hot-spare disk (disk 3) will keep the data as ad, be, cf. Once one failed (lets assume disk 1) then the hot-spare will take the place and keep the data as a, b and c but not d, e or f, then when anyone of these two failed at the same time parity breaks, so the system crashes. This is simultaneous failure.
        But if one drive got replaced and synced properly and then another HDD fails it will definitely recover. Which is not simultaneous failure.

  22. Raid 5, 3 disks, 1 spare

    Disk 1 reporting failure imminent
    Disk 2 3 okay
    Disk 4 SPARE

    I have another disk ready to replace the DISK1.

    ML350 G4.

    How do I safely ‘crash’ the raid to make Disk 4 Spare take over from Disk 1?

    I have read that you need PERC6 to be able to re-insert the new drive in Bay1 and have the SPARE copy the data back.

    For now, I just want to know how to safely crash the array; or is there a better way?

    Thanks all

    1. Found my answer:
      If you insert a hot-pluggable drive into a drive bay while the system power is on, all disk activity in the array pauses for a second or two while the new drive is spinning up. When the drive has achieved its normal spin rate, data recovery to the replacement drive begins automatically (as indicated by the blinking Online/Activity LED on the replacement drive) if the array is in a fault-tolerant configuration.
      If you replace a drive belonging to a fault-tolerant configuration while the system power is off, a POST message appears when the system is next powered up. This message prompts you to press the F1 key to start automatic data recovery. If you do not enable automatic data recovery, the logical volume remains in a ready-to-recover condition and the same POST message appears whenever the system is restarted.

  23. I have a RAID Level 5 Box and I have installed Microsoft Dynamics GP on it. Hence it is a database server. I need to have a full IMAGE copy so incase of any disaster I can instantly recover. Can anyone tell me how I can take GHOST/IMAGE of all RAID 5 machine? And how I can restore it?

    Any help would be highly appreciated.

    1. RAID doesnt care what is on the disks. It looks at them as just ‘Disks’. In RAID5 when a disk fails, it can recover. So you do not Ghost RAID, you Ghost DATA. If that data corrupts, you restore it from a BACKUP/IMAGE. If a DRIVE fails. RAID recovers it for you. I have become a fan of RAID5 with a spare now. A HOTSPARE is an option in most RAID setup choices.
      The set-up is 1+1+1 = RAID5
      1 = HOT SPARE

      RAID5 protects the DRIVES (and the data they contain) against a failure. Having the HOT SPARE allows the RAID utility to fall back onto the SPARE upon drive failure, therefore your RAID is again back at 3 drives and again can lose 1 drive without data loss.

  24. We have 1 server from fujitsu siemens(OS Windows 2003 server) with raid5(6HDD,73GB each) but no hotspare drive. One HDD suddenly found damage. In this situation if we want to rebuild the array, should we just put the same size new HDD in there? After inserting the new HDD what will happen at the beginning? Will the rebuild start automatically or we have to do anything in the Web Bios.

  25. I am little confused. Please help. I have 7 hard disk configured with raid 5. I don’t have any Hot spare drive. What will happen if 2 disk are failed at a time. If all the data lost why is the best solution for redundant …?

    1. hi Jack,

      it does not matter how many drives you have in Raid 5 (only 1 drive can fail) if 2 drives fail then it is difficuilt to recover the data so you will loose data.

      for redundancy..
      Raid 6 can have max 2 disk failures.
      or Raid 10 can have 2 disk failures in different mirror sets.

  26. In the raid 5 features, u have mentioned “hot spare option supported”, so how does it mean?
    1) is it means that it will take a spare drive automatimaclly in the same logical pool when it detects a drive failure or no need for human intervention?
    2) is it we need to configure the spare drive to that array?

  27. Dear sir/Mam

    This is Madan already i ve configured raid 5 in dell server each hard disk is 1 Tb i installed os win server 2008 r2 on disk 0 while i m removing the disk 0 why at that time my sys is getting restart?

  28. I have raid 5 with 3 hard drives. One of the hard disk failed in box 1. We did have a spare hard drive and we had to source for a replacement. In the meantime there was a prolonged power failure and the server initiated an automatic shutdown. During the time the server was off. We removed the failed hard drive. Will the server boot with the remaining hard drive as we await to replace the fautly one?. Anybody with a solution.

    1. Will the server boot with the remaining hard drive as we await to replace the fautly drive? Yes, but if another drive fails you will lose ALL data. I suggest you back up the data on the system ASAP .

  29. i have raid 5 and i know that if two hard drive fails, then my data are lost, so how should i retrieve my data after replacing with new hard drive?.

  30. Dumb question here. My 3 HD 18g. ea. RAID5 array is degraded due to a failing hard drive.
    I have another drive to replace it with, but this drive may have data on it. It came out of another deactivated RAID5 server. If I pull the bad drive out, plug this other drive in…am I going to have a problem or will it rebuild to the existing RAID5 config?

  31. If 1 of the Disks fails in RAID5, can i do I/O operations on that LUN(RAID 5 LUN). Other i need to wait it it rebuild? if one disk fails what will be the status of my LUN. i mean like failed or degraded or etc…

  32. If the Status of the RAID 5 is Degraded, can i do Read write operations on it?

    If yes please explain how.

    If No, then how is support a fault tolerance?

    Thanks in Advance…

  33. evn aftr buyng a new hdd, my cpu is unable to detct it. nothng is being shown on monitr and lights r nt glowing in mouse n keyboard. plz hlp me diagnose the problem

  34. i have 28 hard disk with raid technology and with 02 hard disk drive failure the system still working. further if i remove three and inserted the same in different slot it will work? and if i replace these three with new one is system work or not?

  35. I’m working on a Dell PowerEdge 2800 with RAID5. one of the disks failed and I replace it with a new identical one. power up the server and the light on it still amber after 24 hours. installed open managed an it still shows the status of the drive as “failed”. Did I need to do something else? I thought the drive would be rebuild automatically?
    Any ideas?

  36. I have a server HP DL380 G8 and I have two hard disk configured as a mirror . one of the hard disk was faild , but still the server is working fine in one hard disk , I add a new hard disk with the same size but still I cannot see the new hard disk my be need to add under the Mirror , my Q is its possible boot from smart CD and add new hard disk under raid configuration without losing the system configuration. ?

  37. 2 days back i had a problem server hard disk and server won’t boot. I have IBM system X3650 series server with 3 x 73.4 GB SAS 3.5″ hard disk with Raid5. After the weekend holiday when i came to office server was down and hard disk 0 is in critical. i tried to reboot unfortunately it won’t. i bought the same size and modelled hard disk and insert the same to the server. Initially it was giving error and i went through the IBM serveraid 8k controller and it rebuilds. After rebuilds the restarted and giving the error A DISK READ ERROR OCCURED PRESS CTL + ALT+ DEL TO RESTART. Keeps on showing the same message on each restarts and server not boot.

    Please guide me what should i do?

  38. Hi Guyz, plz I have dl 380 g8 server that have 8TB HDD on bay1 and 8TB HDD on bay 2. I used RAID 1 for OS (1st and 2nd slot) the remaining 6TB HDD on bay 1, I used RAID 5. Guy, which appropriate raid can I use for 8TB HDD at the 2nd bay to achieve 14TB HDD. (6TB bay 1 plus 8TB HDD bay 2)?


  39. i have problem with my xitrix, raid 5, i had 12 pcs HDD with 3TB. my problem is the first HDD is damage… can i still replace this 1st HDD and back to normal or still needed for any settings or configuration to fixed this problem? my OS is windows 2008 server. we used hardware raid., my 2nd question is some of the opinion i get is that they said the OS is being installed at 1st HDD. :( your feedback is highly appreciated..



  40. Hello guys.

    I know that this Q may have been asked before… :-)

    Imagine that i have a BOX with 9 300GB 10K SAS drives.

    ||| __ __ __ __ __ __ __ __ __ |||
    ||| 1 2 3 4 5 6 7 8 9 |||
    |———————————–| HS
    RAID 5

    Disk 1-8 runs RAID5 and disk 9 is hot-spare… Disk 4 dies and the hot-spare jumps in and is bieng rebuild as disk 4… Now disk 4 is “By-passed” by IBM DS Storage Manager and DSM says that i cannot take out disk 4… Then on a forum, people writes that i can do something like “Force fail disk” i know this feature, but i can’t use it…

    What the hell do a wise one do here…?

    If i kill the hot-spare disk, then i can “Force Fail” disk 4 and replace it, but if any other disk failes when the rebuild is running, then i’m skrewed…

    1. my question is if i have server 2008 and my server got corrupt means.Now Window server hard disk not working.Its got damage system not getting ON now how i can take the my Active directory backup.Which i have its a old backup.So guys tell me now what i can do for AD backu.rember system not getting on means HD got damage.every think i did with hd repair and all that still its in same condition now rply the ans….plzzzzzzz

  41. I have IBM Storwize V3700 with one failed drive in RAID 5 array. We don’t have spares to replace the drive immediately. Is it ok to let storage run with one failed drive or is there a possibility that parity is overwritten if I let users to use the storage?

    Also, does replacement drive need to have EXACTLY the same firmware level as the original? Or is it enough that model and size are the same?

    Still, have a question? Get help on our forum!