1 Reply Latest reply: Aug 12, 2014 11:30 PM by Nick Dyer RSS

    Controller Failover Events

    Varghese Pulipra Cherian Wayfarer


      Hi Folks,

       

      I  wanted to clarify   the  following.

       

      a. If reboot of the active controller would trigger failover ?

      b. What would happen  to the  raid rebuild  during controller failover would it continue  ?

      c. If we have  dual disk failure at once what  would happen , would the array  shutdown immediately ?

       

      regards

      Varghese

        • Re: Controller Failover Events
          Nick Dyer Navigator

          Hello Varghese,

           

          A - Typically yes, if the active controller needs to reboot for whatever reason (the usual culprit is a firmware upgrade) then the standby controller will take the active role and will do so with no performance impact. It's rare that the active controller needs to reboot outside of a firmware upgrade.

           

          B - Yes this would continue, as both controllers are a mirrored pair for eachother.

           

          C - Before NimbleOS 2.1.4 the array runs RAID 6 + Hotspare, meaning it's possible to lose two drives in the RAID set on an individual drive shelf with no data loss. However, losing a third drive in the RAID set will result in all data being corrupted. Therefore if two drives fail at the same time the array will halt data services to ensure data integrity, as if two drives have failed there's a good chance a third failure may be on it's way. The array is still online during this process, it's just data is not being served.

           

          In NimbleOS 2.1.4 we have moved to RAID-3P (Triple Parity RAID) which allows for three simultaneous drive failures in a single drive shelf before we have to take any corrective action. I blogged about it here: Nimble OS 2.1 Part 5 - Introduction To RAID-3P (Nimble's New Triple Parity RAID Implementation)

           

          HTH!