This Knowledgebase article provides general information about upgrading hardware and software on servers with Neverfail installed.
While this Knowledgebase article provides general information, for specific procedures about upgrading hardware and software with Neverfail installed, please refer the appropriate Knowledgebase articles.
Upgrading Systems Without Downtime
Neverfail provides high availability minimizing the need for downtime during maintenance tasks. Due to their individual nature, some maintenance tasks are better suited to this process than others are.
Neverfail treats both servers equally allowing the users to operate from either without the need to stop both servers to perform a restore. When connected, the active server will synchronize and replicate to the passive.
When considering maintenance tasks it is important to understand what the maintenance task is doing in order to determine the best approach when Neverfail is installed.
Neverfail can continue to provide access to one server (the active server) in the Neverfail server pair while the other server (the passive server) is undergoing a hardware upgrade. Once the upgrade of the passive server is complete, simply initiate a switchover to perform the upgrade on the previously active (now passive) server while users continue to operate.
The following should be used as a guideline only as this is not a comprehensive step-by-step procedure on how to perform a hardware upgrade.
- Switchover the server pair (if necessary) so the server to be worked on is passive.
- Shut down Neverfail Heartbeat leaving the protected applications running.
Take the passive server off-line and replace appropriate hardware.
Hardware upgrade considerations:
- System Disk replacement may require a restore.
- NIC replacement may require the Packet Filter to be reinstalled.
- Changes to Primary Hardware may require a new License key.
- Once complete, re-establish the pair and allow to resync.
- Once synchronized, switchover if required.
Software upgrades may be a little more complicated as they tend to vary in action. Prior to performing an upgrade, the following should be considered:
- What Neverfail is protecting (Registry and Data)
- What the Update is updating (Registry Application, Data & Data Structure)
With these in mind, the most appropriate upgrade process can be performed.
- A Maintenance task that only adds files into the protected file structure can be performed on the active server. The changes would be replicated and no down time would be required.
- A maintenance task that updates the registry and or application files, which are not part of the replication set, should be applied to both servers. This operation can usually be carried out during normal operation. Servers can be switched before a reboot should one be required for client access to avoid downtime.
Note: If the server needs to be rebooted following an update, the replication must complete and show as being fully synchronized before a switchback is initiated. Once this has completed, it is possible to reboot the other server following the maintenance.
Upgrades That Change Applications and Data Structure
Some maintenance processes change the application and the structure of a database. In such cases, care must be taken when Neverfail is running because a mismatch between application level and data structure can occur. With Heartbeat running, changes to the data structure will be replicated, however the application files and registry changes will not. Following a switchover, the original application will start to work on the updated data structure, which may compromise data integrity. In such cases, switching between servers to avoid down time is inappropriate.
Updates That Require Access to Network Resources
Some applications require access to network resources such as Active Directory or DNS for upgrade purposes. In these circumstances, the upgrade can only be performed on the Active server.
The active server can be switched in order to keep users connected to their applications during maintenance tasks. The process of each maintenance task needs to be considered on its own merits to determine the suitability and associated risk. One role of a highly available system during maintenance is to provide the ability to quickly roll back should that maintenance task fail. A passive server can quickly be brought online following a failed update allowing technicians to troubleshoot the problem without significant impact to the business.