An online migration using commands like ‘hagrp -online -sys ‘ moves the service group and its resources to another cluster node without service interruption. This allows the original node to be upgraded while the application remains running on the secondary node, after which the group can be migrated back. This is a standard procedure for planned maintenance with high availability.

An AutoStart value of 2 means ‘Failover’ mode. The service group will start automatically on the node where it was last running, provided that node is booting and is a valid member of the cluster. If that node is unavailable, it will not automatically start on a different node, which is a key distinction from the ‘Global’ (value 1) setting and a common point of confusion in troubleshooting.

When a service group is FAULTED, you must first clear the faulted state using `hagrp -clear`. This command resets the group’s state machine. Only after clearing the fault should you attempt to bring the group back online with `hagrp -online`. Attempting to switch or online the group without clearing the fault first will fail or cause unexpected behavior.

Veritas Volume Replicator (VVR) provides host-based, real-time, synchronous replication of volumes. When paired with array replication, it offers a robust solution for critical tiers like databases, enabling near-zero RPO (no data loss) by ensuring data is written to the remote site before acknowledging the write to the application. VCS Geographic Edition manages the orchestration of failover but relies on underlying replication technologies like VVR for the data copy.

The ‘JOINING’ state is a transient phase where a node is communicating with the cluster and receiving the latest cluster configuration (lltconfig, gabconfig, main.cf). It appears after a node boots or when the VCS engine is started. It indicates the node is not yet a fully active, voting member but is in the process of becoming one.

Dependencies between resources are defined in the `Dependency` tree section of the service group in the main.cf file. This is a distinct, hierarchical section that defines parent-child relationships (e.g., `Application_Resource requires Mount_Resource`). The cluster uses this tree to determine the correct order for online and offline operations, not attributes within individual resource blocks.

Veritas Storage Foundation Storage Checkpoints (VxSC) are designed for this purpose. They can be integrated with application scripts: a ‘pre-script’ to quiesce the database (put it in backup mode), a ‘freeze’ to create the consistent checkpoint, and a ‘post-script’ to unquiesce the database. This ensures data consistency with minimal I/O interruption, making it far more efficient and reliable than breaking mirrors or using non-application-aware snapshots.

The order of nodes in the `NodeList` attribute defines the preferred order for the service group. The `FailOverPolicy` attribute (e.g., `CONFIGURED_NODE`) then dictates how the cluster uses this list. Setting `FailOverPolicy` to `CONFIGURED_NODE` tells VCS to fail the group over to the next available node in the `NodeList` sequence, enforcing a predictable failover path rather than a random selection.

The `lltstat -n` (node) command displays Low Latency Transport statistics for heartbeat traffic between nodes. It shows sent/received packets, errors, and transport state (e.g., `OPEN` or `DOWN`) for each link, providing direct insight into the health and performance of the private heartbeat network. `gabconfig -a` shows group membership but not the underlying link statistics.

An online migration using commands like ‘hagrp -online -sys ‘ moves the service group and its resources to another cluster node without service interruption. This allows the original node to be upgraded while the application remains running on the secondary node, after which the group can be migrated back. This is a standard procedure for planned maintenance with high availability.0