none
Prevent/Disable "Redirected access" RRS feed

  • Question

  • hi Guy,

    Windows Server 2016, Failover cluster, fiber connect to storage, I get a problem that if fiber connection disconnected on node-1, the CSV storage auto enabled Redirected Access, instead of failover the hyper-V guest to node-2.

    That seriously impact the VM guest performance.

    It can't turn off redirected access in GUI, it need restart cluster service in both node.

    May I ask how to prevent redirected access auto enabled/turn on upon fiber cable disconnected? I search in Internet can't find any solution that can prevent redirected access enable, it is very annoying.

    Regards

    Thanks...KEN

    Thursday, December 5, 2019 11:15 AM

All replies

  • Hi KEN,

    >>That seriously impact the VM guest performance.

    >> the CSV storage auto enabled Redirected Access, instead of failover the hyper-V guest to node-2.

    Yes. This might cause slower performance since we are no longer using direct local block level to storage.

    Instead, we are using remote file system access via the network using SMB.

    When a storage connectivity failure for a node, then CSV will be in a Redirected Access mode.

    Mostly, the following events will lead to failover:

    • Host crash
    • Power failure
    • Network Isolation
    • Monitored service failure

    As far as I know, this behavior is by design and there is no way to disable Redirected access when the storage connection disconnected.

    For your reference:

    Troubleshooting ‘Redirected Access’ on a Cluster Shared Volume (CSV)

    Best Regards,

    Candy




    Please remember to mark the replies as an answers if they help.
    If you have feedback for TechNet Subscriber Support, contact tnmff@microsoft.com   


    Friday, December 6, 2019 7:49 AM
  • Hi Candy,

    Thanks for your reply, but consider there have several hyper-V vm guest running on the SAM disk, using network instead fiber and it won't trigger cluster failover is not prefer.

    I am remember when it single NIC cable plug, it can trigger cluster resource failover, but after I configure teaming with two NIC cable, it wont' trigger cluster resource failover, and use Redirected Access instead.

    Is there have fine tuning for that? Use failover instead of Redirected Access?

    Regards

    Thanks...KEN

    Sunday, December 8, 2019 3:58 AM
  • Do you have MPIO configured for your fibre connection?  You need to have two paths to the storage in order for failover to occur to the other channel.  If you have your fibre configured with just single connections, no MPIO, then you have configured single points of failure.  If each host has just a single connection to the storage, then the only way to continue operations is to use redirected IO through a host which has not lost connection to the storage.

    tim

    Sunday, December 8, 2019 2:29 PM
  • Hi Tim,

    Yes, the two node cluster has MPIO that echo node server have two fiber connections to SAN storage, my client query that if all fiber connection disconnected in a active node server, how come it won't trigger resource failover, but using Redirected Access that degraded the Hyper-V guest performance.

    If this is Microsoft working as design and there is "NO" way to modify that, do it have Microsoft official link to mention that? so that I can reply to client that "Nothing" we can do.


    Regards

    Thanks...KEN

    Monday, December 9, 2019 1:55 AM
  • Hi ,

    >>my client query that if all fiber connection disconnected in a active node server, how come it won't trigger resource failover, but using Redirected Access that degraded the Hyper-V guest performance.

    Based on our experience, the cluster resource will not failover to other node when all path to the storage in a active node server are disconnected.

    As I said before, mostly, the following events will lead to failover:
    • Host crash
    • Power failure
    • Network Isolation
    • Monitored service failure

    Storage disconnection on one node will not lead to cluster resource failover.

    I use ISCSI storage to do the test and the result is the same. The IO traffic will get redirected over network to the other node.

    Best Regards,

    Candy


    Please remember to mark the replies as an answers if they help.
    If you have feedback for TechNet Subscriber Support, contact tnmff@microsoft.com   


    Monday, December 9, 2019 3:25 AM
  • Hi KEN,

    Just want to confirm the current situations.

    Please feel free to let us know if you need further assistance.                   

    Best Regards,

    Candy


    Please remember to mark the replies as an answers if they help.
    If you have feedback for TechNet Subscriber Support, contact tnmff@microsoft.com   

    Wednesday, December 11, 2019 1:30 AM
  • hi Candy,

    network failure and host fail would trigger the hyper-V VM migration, but it need wait for 3 minutes, but to File Server resource, it is very slow.

    Regards

    Thanks...KEN

    Thursday, December 12, 2019 12:31 PM