HP HSG80 manual Pause During Fabric Reconfiguration, Controller Saturation

Common Platform Issues

Using Switch Zoning to Prevent Crashes When the 96-Connection Limit Is Exceeded

The HSG80 controller may crash if more than 96 connections exist on the fabric. HP suggests that you use switch zoning to limit the number of connections visible to a single HSG80 controller. For more information about switch zoning, refer to the HP StorageWorks Data Replication Manager HSG80 ACS Version 8.7P Configuration Guide.

I/O Pause During Fabric Reconfiguration

During a fabric reconfiguration, you will notice a brief pause in I/O functions on all servers connected to the fabric. This brief cessation of read/write operations is normal. A fabric reconfiguration can be caused by switches starting up or shutting down, and by the physical plugging or unplugging of fiber cables.

Controller Saturation

High usage of many remote copy set and nonremote copy set LUNs has a serious impact on the performance of the remote copy set LUNs, causing controller saturation and possibly starving a full copy operation.

A saturated controller condition begins approximately when idle time falls to 25% or less when viewed through VTDPY. When this occurs, you may see Aborted Command errors through the Command Line Interface (CLI).

To prevent controller saturation:

■Avoid placing multiple heavy-use loads on the controllers.

■Use the VTDPY screen to monitor controller idle time percentage. Adjust load to maintain an idle time of 25% or greater.

RAID 5 Remote Copy Set Target Drop

Under the following combined conditions, RAID 5 remote copy set targets may be dropped:

■No write history log disk is configured, and

■Host I/O is accessing the initiator remote copy set LUNs.

When both target controllers are shut down, a full copy operation is triggered. When the target controllers are restarted, they will begin a 3-minute memory diagnostic.

While the memory diagnostics are running, the full copy I/O to RAID 5 target LUNs is stalled. Therefore, after approximately 2 minutes, a timeout occurs and the target LUNs are dropped from the remote copy set. This target drop problem occurs only on RAID 5 remote copy set LUNs.

■To prevent target LUN drop:

1.Set port_2_topology on both initiator controllers to offline before target controllers are booted and powered on.

2.Wait five minutes after target controllers are restarted to allow for memory diagnostics to complete.

3.Set port_2_topology on the initiator controllers back to fabric.

Data Replication Manager HSG80 ACS Version 8.7P Release Notes 9