Servicing storage systems

Storage systems that are supported for attachment to the SAN Volume Controller are designed with redundant components and access paths to enable concurrent maintenance. Hosts have continuous access to their data during component failure and replacement.

The following guidelines apply to all storage systems that are attached to the SAN Volume Controller:

Always follow the service instructions that are provided in the documentation for your storage system.
Ensure that there are no unfixed errors in the SAN Volume Controller error log before you perform any maintenance procedures.
After you perform a maintenance procedure, check the SAN Volume Controller error log and fix any errors. Expect to see the following types of errors:
- MDisk error recovery procedures (ERPs)
- Reduced paths

The following categories represent the types of service actions for storage systems:

Controller code upgrade
Field replaceable unit (FRU) replacement

Controller code upgrade

Ensure that you are familiar with the following guidelines for upgrading controller code:

Check to see if the SAN Volume Controller supports concurrent maintenance for your storage system.
Allow the storage system to coordinate the entire upgrade process.
If it is not possible to allow the storage system to coordinate the entire upgrade process, perform the following steps:
1. Reduce the storage system workload by 50%.
2. Use the configuration tools for the storage system to manually failover all logical units (LUs) from the controller that you want to upgrade.
3. Upgrade the controller code.
4. Restart the controller.
5. Manually failback the LUs to their original controller.
6. Repeat for all controllers.

FRU replacement

Ensure that you are familiar with the following guidelines for replacing FRUs:

If the component that you want to replace is directly in the host-side data path (for example, cable, fibre-channel port, or controller), disable the external data paths to prepare for upgrade. To disable external data paths, disconnect or disable the appropriate ports on the fabric switch. The SAN Volume Controller ERPs reroute access over the alternate path.
If the component that you want to replace is in the internal data path (for example, cache, or disk drive) and did not completely fail, ensure that the data is backed up before you attempt to replace the component.
If the component that you want to replace is not in the data path, (for example, uninterruptible power supply units, fans, or batteries) the component is generally dual-redundant and can be replaced without additional steps.