API Doc
Search Docs...
⌘ K
OverviewDeploymentManagementOperationReferenceGlossary
    ACOS 6.3.0
  • Acrfra Cloud Operation System cluster>
  • Failure scenarios>
  • SCVM failures

Single SCVM failures

  • I/O rerouting enabled

    I/O rerouting can detect the health status of the SCVM. If the local SCVM fails, the destination address of I/O requests from local virtual machines will be modified from the local SCVM to a remote SCVM in the ESXi routing table. As a result, I/O requests from local virtual machines are rerouted to an SCVM on another ESXi host, completing the failover of the faulty path. However, when the local SCVM recovers to a normal state for more than 3 minutes, I/O requests are rerouted to the local SCVM, and the temporary routing path is cleared. The switching process completes within milliseconds, usually without causing any noticeable I/O interruption.

    In addition, if the failed SCVM was running ZooKeeper and Meta services, the failure will trigger re-election of the ZooKeeper Leader and Meta Leader, during which a brief I/O interruption may occur.

  • I/O rerouting not enabled

    When I/O rerouting is disabled, if virtual machines are configured with HA, an SCVM failure will trigger HA for the virtual machines on the host where the SCVM resides. Otherwise, the virtual machines will be unable to perform any read or write operations.