Subscribe to this information
POWER7 information
The planning checklists contain a list of tasks that must be completed before you perform CEC hot-node add and hot-node repair maintenance operations on IBM® Power® ESE (8412-EAD), IBM Power 770 (9117-MMB, 9117-MMC, and 9117-MMD) and IBM Power 780 (9179-MHB, 9179-MHC, and 9179-MHD) systems. It also lists which tasks are associated with each type of operation.
For guidelines that apply to all CEC adapter hot-add and hot-repair maintenance operations, see Prerequisites for CEC hot-add and hot-repair. These general guidelines and the following checklist must be followed.
| Operations | Model 8412-EAD, 9117-MMB, 9117-MMC, 9117-MMD, 9179-MHB, 9179-MHC, and 9179-MHD tasks | Responsibility |
|---|---|---|
Hot-node add |
Ensure that sufficient space is available in the rack that contains the system for another node to be installed under the existing nodes. |
Customer |
Hot-node add |
Adding a node increases the power requirements of the system. If necessary, the power circuits that supply the system must be upgraded to handle these additional power requirements. This upgrade must be complete before the hot-node add begins. The power requirements of additional nodes are detailed in Chapter 4, Server Specifications and Chapter 8, Planning for Power in Site and Hardware Planning. |
Customer responsibility to perform the required site planning activities |
Hot-node add, hot-node upgrade (memory), hot-node repair, concurrent GX adapter add, and hot GX adapter repair |
If the system is managed by a Hardware Management Console (HMC), verify that the HMC level is the same as, or later than, the level shown in Model 8412-EAD system firmware and HMC levels, Model 9117-MMB and 9179-MHB system firmware and HMC levels, Model 9117-MMC and 9179-MHC system firmware and HMC levels, or Model 9117-MMD and 9179-MHD system firmware and HMC levels. Verify that the system firmware level is the same as, or later than, the level shown in Model 8412-EAD system firmware and HMC levels, Model 9117-MMB and 9179-MHB system firmware and HMC levels, Model 9117-MMC and 9179-MHC system firmware and HMC levels, or Model 9117-MMD and 9179-MHD system firmware and HMC levels. For information about determining firmware levels using the Hardware Management Console (HMC), see Using the HMC to view the existing firmware levels. For information about disconnecting an HMC from a managed system, see Removing HMC connection data. |
Customer Service representative must verify |
Hot-node upgrade (memory), hot-node add, and hot-node repair |
CAUTION: Observe the following with the Symmetric Multiprocessing (SMP) and the service processor cables on IBM Power ESE, Power 770, and Power 780 systems when performing a CEC hot add or repair operation:
|
Service representative must verify |
Hot-node upgrade (memory), hot-node repair, and hot GX adapter repair |
If there is only one partition running on the system, the system can be powered down if that partition must be shut down during the CEC hot add or repair operation. To verify or change the setting that controls this behavior, see System power off policy in Powering the system on and off. |
Service representative |
Hot-node add, hot-node upgrade (memory), hot-node repair, concurrent GX adapter add, and hot GX adapter repair |
Verify that service processor redundancy is enabled, except on a system with a single node. To enable service processor redundancy on the managed system on an HMC, in the navigation area, select . Failover status is displayed in the contents area and can be enabled or disabled. Note: The server must be power
cycled for the enabling of service processor redundancy to take effect.
If power cycling is required, the node can be added nonconcurrently
(with power off).
For information about redundancy on the HMC, see FSP Failover. |
Customer Service representative must verify |
Hot-node add, hot-node upgrade (memory), hot-node repair, concurrent GX adapter add, and hot GX adapter repair |
If the system is managed by an HMC, verify the network connections between the system service processors and the HMC before the service representative arrives by using the service processor status function on the HMC. For information about testing the network connections on the HMC, see Testing the connection between the HMC and the managed system. |
Customer Service representative must verify |
Hot-node add, hot-node upgrade (memory), hot-node repair, concurrent GX adapter add, and hot GX adapter repair |
Ensure that Electronic Service Agent™ (ESA) is enabled to call home. For additional details about the call home function on the HMC, see Manage Systems Call-Home. |
Customer Service representative must verify |
Hot-node add, hot-node upgrade (memory), hot-node repair, and hot GX adapter repair |
For 9117-MMB and 9179-MHB systems, you should quiesce the server before the hot-node operation. This means that all critical application should be halted or moved to another server using Live Partition Mobility (LPM) before the operation begins. Noncritical applications can be left running. The partitions can be left running at the operating system command prompt. For 8412-EAD, 9117-MMC, 9117-MMD, 9179-MHC, and 9179-MHD systems, you must quiesce the server before the hot-node operation. This means that all critical applications must be halted or moved to another server using Live Partition Mobility (LPM) before the operation begins. Noncritical applications can be left running. The partitions can be left running at the operating system command prompt. For additional information on a system managed by an HMC, see Mobility in the Systems management for partitions topic in Managing the Hardware Management Console. Note: IBM recommends that adequate protection is in
place for business critical applications before you perform CEC hot-node
add and hot-node repair maintenance operations. To mitigate the possibility of an unexpected
outage, you can manually move critical workloads to another server
before the operation begins. Performing CEC hot-node add and hot-node
repair maintenance operations during a regularly scheduled maintenance
window or during off-peak hours reduces the impact of an unplanned
outage.
|
Customer |
Hot-node add and hot-node upgrade (memory) |
Obtain and apply Capacity Upgrade on Demand (CUoD) activation codes prior to the operation so that the memory resources are available immediately after the hot add operation. For information about CUoD activation codes, see Capacity Upgrade on Demand activation codes. |
Customer |
Hot-node add and hot-node upgrade (memory) |
If the customer wants to dynamically add the new processors and memory resources to activated partitions after the Hot-node add, the profiles that were used to activate the partitions must have the maximums for memory set to the appropriate values. When the Hot-node add is complete, the processor and memory resources can be added to the partitions by using dynamic logical partitions (DLPAR). For information about DLPAR, see Dynamic logical partitioning. |
Customer |
Hot-node upgrade (memory) |
Verify that more than one node is present in the system. Hot-node upgrade (memory) cannot be done if the system has only one node. |
Customer Service representative must verify |
Hot-node repair, hot GX adapter repair, and hot-node upgrade (memory) |
Run the Preparing for Hot Repair or Upgrade utility on the HMC to determine if the system is prepared for a node upgrade (memory). For more information about the utility, see Preparing for hot repair or upgrade. |
Customer Service representative must verify |
Hot-node add and concurrent GX adapter add |
System firmware enforces the node and GX adapter plugging order. Only the next GX adapter slot or node position based on the plugging order is available. |
Customer Service representative must verify |
Concurrent GX adapter add |
Ensure that there is a GX adapter slot reservation for each GX adapter that is being added. Each installed GX adapter and GX adapter slot reservation consumes approximately 128 MB of memory. Additional GX adapter slots can be reserved by using the Advanced System Management Interface (ASMI). New GX adapter slot reservations take effect the next time the system is powered on. |
Customer |
Concurrent GX adapter add |
The GX adapter cannot be added to a node that has been deconfigured. The node must be repaired and reconfigured before another GX adapter can be added. To verify that all nodes in the system are configured, see Using the ASMI to view deconfigured resources. |
Customer Service representative must verify onsite1 |
Hot-node add |
When multiple hot adds are planned that include node and GX adapter adds, the GX adapter(s) should be installed in the node before the node is installed in the system. |
Service representative |
Hot-node add and concurrent GX adapter add |
For multiple upgrades that include node or GX adapter adds, as well as I/O drawer adds, the node or GX adapter add must be completed first. The I/O drawer can then be added as a separate concurrent I/O drawer add. |
Service representative |
Hot-node repair and hot GX adapter repair |
Ensure that the replacement parts are the same type as the parts that are being removed. |
Service representative |
Hot-node repair and hot GX adapter repair |
Replace only the existing parts. New hardware must not be added during a Hot-node repair or a hot GX adapter repair. |
Service representative |
Hot-node repair |
Ensure that existing node hardware is not moved during the Hot-node repair. |
Service representative |
1 Hardware in a node can fail at any time. When this happens a node can become unconfigured during system IPL to isolate the failure. The failure and unconfigured event can occur after the hardware upgrade package has arrived at the customer site and the service representative has been dispatched. The service representative must ask the customer if the system has any outstanding deferred repairs. Note: The hot-node
add function adds and enables the I/O adapters only in the new node.
|
||
