Recovering from an orchestrator system failure in Metro Mirror configurations

Edit online

To recover from an IBM Spectrum Virtualize 3-Site Orchestrator system failure in a Metro Mirror configuration, note the following considerations.

  • If an orchestrator system fails, orchestrator services automatically restart and reconstruct the configuration based on the existing configuration.
  • If IBM Spectrum Virtualize 3-Site Orchestrator failed after the creation of a 3-site configuration, the configuration is automatically recovered based on the availability of all three sites. Enter the ls3siteconfig command to verify that the site is not available.
  • If all three sites are accessible and all the objects of the consistency group are recovered correctly, data cycling is automatically restarted in the 3site_consistent state.
  • If all three sites are accessible and all the objects of the consistency groups are recovered correctly, but the periodic source of the configuration cannot be identified, the state of the 3-site consistency group is stopped.

    If you choose to, you can select the primary site as the periodic source that uses the ch3siteconsistgrp -periodicsrc command. Selecting the primary site as the periodic source starts the data cycling in 2site_periodic mode.

  • If the near site is unavailable and the recovered configuration is a star topology, data cycling automatically starts in 2site_periodic mode.
  • If the near site is unavailable and the recovered configuration is a cascade topology when IBM Spectrum Virtualize 3-Site Orchestrator failed, the state of the 3-site consistency group is stopped.

    If you choose to, you can select the primary site as the periodic source by using the ch3siteconsistgrp -periodicsrc command. This command starts the data cycling in 2site_periodic mode.

  • In scenarios in which the 3-site objects are partially recovered, the state of the consistency group is partial.
  • If the primary or auxiliary-far site is unavailable, the state of the consistency group is stopped after the orchestrator recovery.
  • On detection that a formerly unreachable site becomes available, IBM Spectrum Virtualize 3-Site Orchestrator automatically restarts to recover the complete consistency group, including the missing site in the configuration and attempts to start data cycling. All other recovery considerations are applicable.
  • If more than one site is unavailable, the orchestrator recovery process fails.
  • If you enter the convertrcrelationship or convertrcconsistgrp commands after an orchestrator system failure, a partial state might result.
    • If the consistency group is in partial state after you issue the convertrcrelationship command, you can continue to convert it to a 3-site configuration, or you can convert it back to a 2-site consistency group by using the convertrcconsistgrp command.
    • If the consistency group is in the partial state after you issue the convertrcconsistgrp command, you can convert it only to a 2-site consistency group.