Veritas InfoScale™ 7.4.1 Virtualization Guide - Solaris

• Online and offline operations for service groups in the StorageSG attribute

  To manually bring online or take offline service groups that are configured in the StorageSG attribute do not use the AlternateIO resource or its service group

  Instead, use service groups configured in the StorageSG attribute.

• Freeze the service group for the AlternateIO resource

  Freeze the AlternateIO service group before you bring online or take offline service groups configured in the StorageSG attribute of the AlternateIO resource. If you do not freeze the service group, the behavior of the Logical Domain is unknown as it is dependent on the AlternateIO service group.

• Configuring preonline trigger for storage service groups

  You must configure preonline trigger in the following scenario:

  When the service groups configured in the StorageSG attribute of the AlternateIO resource are of fail over type, and if you accidentally bring storage service groups online on another physical system in the cluster.

  It is possible to bring the storage service groups online on another physical system because resources configured to monitor back-end storage services are present in different service groups on each physical system. Thus, VCS cannot prevent resources coming online on multiple systems. This may cause data corruption.

  Note:

  Perform this procedure for storage service groups on each node.

  To configure preonline trigger for each service group listed in the StorageSG attribute

  • Run the following commands:

    # hagrp -modify stg-sg TriggerPath bin/AlternateIO/StorageSG
    # hagrp -modify stg-sg TriggersEnabled PREONLINE

    where stg-sg is the name of the storage service group

• Set connection time out period for virtual disks

  When a disk device is not available, I/O services from the guest domain to the virtual disks are blocked.

  Veritas recommends to set a connection time out period for each virtual disk so that applications times out after the set period instead of waiting indefinitely. Run the following command:

  # ldm add-vdisk timeout=seconds disk_name \
  volume_name@service_name ldom

• Fail over of LDom service group when all the I/O domains are down. When the attribute SysDownPolicy is set to AutoDisableNoOffline for a service group, the service group state would be changed to OFFLINE|AutoDisabled when the system on which the service group online goes down. Before auto-enabling the service group and online the service group on any other nodes, it is mandatory to ensure that guest domain is stopped on the system (control domain) that is down. This is particularly important when failure-policy of the master-domains is set to ignore.

If the guest domain need to be made available even when primary domain is rebooted or shut down for maintenance.

• The DomainFailurePolicy attribute would be changed to {primary=ignore, alternate1=stop} or {primary=ignore, alternate1=ignore}.

  Guest domain will not be stopped even when primary domain is rebooted or shutdown

• The SysDownPolicy attribute would be set to AutoDisableNoOffline for planned maintenance. VCS will not fail-over the service group when the node is down instead the group would be put in to auto-disabled state.

When the control domain is under maintenance, and if the alternate I/O domain fails due one of the following:

• DomainFailurePolicy attribute is set to {primary=ignore, alternate1=stop} and only the I/O services from alternate I/O domain are unavailable (i/o domain is active, but n/w or storage loss).

• DomainFailurePolicy attribute is set to {primary=ignore, alternate1=ignore} and if the alternate I/O domain is down (domain is in-active).