Seamless Failover
Introduction
Seamless SMB Failover is an optional feature for Access Zone and IP Pool failover. It helps reduce client disruption during failover by maintaining client connectivity. Clients reconnect automatically to shared storage without needing to remount shares.
This document outlines the requirements, configuration, how the feature works, and where to find related guides.
Requirements
Seamless SMB Failover is available in version 2.12 and later.
If you are using version 2.12 or 2.13, you will need to add the following command to the PowerScale sudoers file:
eyeglass ALL=(ALL) NOPASSWD: /usr/bin/isi_for_array netstat*
-
Failover must be configured for either Access Zones or IP Pools. Refer to the following guides:
-
A working SMB environment is required, along with DNS resolution for client reconnection.
Configuration Steps
In version 2.14.0 and later, Seamless SMB Failover removes and then re-adds pool interfaces after policies fail over. This updated method avoids the need to add netstat permissions for the eyeglass user in the PowerScale sudoers file, unless the previous Seamless Failover method is required.
To use the previous method, which used netstat to identify connected clients and then disconnected each client using the cluster REST API, set the seamless_failover_re-add_interfaces tag in system.xml to false.
How Seamless SMB Failover Works
Seamless SMB Failover manages client reconnection during failover by removing and reassigning network interfaces associated with the affected Access Zone or IP Pool.
-
Network Interface Removal: During failover, network interfaces associated with the affected pool are removed from the source cluster.
-
Network Interface Reassignment: The same network interfaces are brought online on the target cluster as part of the failover process.
Key Behaviors
The following behaviors describe how Seamless SMB Failover operates during Access Zone and IP Pool failover scenarios:
| Scenario | Result |
|---|---|
| SMB share mounted via FQDN | ✅ Supported — mount via FQDN is required |
| Remounting share after failover | ✅ Not required — seamless reconnection |
| File open and edited before failover, closed after failover completes | ✅ Data is saved successfully |
| File closed or newly created while failover is in progress | ⚠️ Data is NOT saved — resulting in data loss |
Best Practices
-
Avoid Creating New Files During Planned Failover
- If a failover event is planned or scheduled, advise users to avoid creating or saving files during the failover window.
-
Ensure Open Files Are Saved After Failover Completes
-
Files that are open and actively edited before failover begins will retain their data, as long as they are saved after the failover process has fully completed. Users should wait for confirmation that the failover is complete before saving critical work.
-
Seamless SMB Failover provides a transparent experience for connected clients, eliminating the need to remount shares. However, any data created or written during the brief cluster transitional (read-only) period may be lost.
-
By following the best practices outlined above—especially avoiding file creation or save operations during failover—users can safely leverage this feature while minimizing the risk of data loss.
-