This chapter describes how to cable together multiple Recovery Appliance racks. It contains the following sections:
See Also:
You can add racks together to increase system capacity and performance. Recovery Appliance has a flexible configuration. A rack has two compute servers and a minimum of three storage servers. A Recovery Appliance X6 or X5 full rack has 18 storage servers and a Recovery Appliance X4 full rack has 14 storage servers. The maximum configuration is 18 full racks.
You can cable together multiple Recovery Appliance racks that have any number of storage servers; they do not have to be full racks. Spine switches connect the InfiniBand network among multiple racks. You must order the spine switches and cables.
Before cabling the racks together:
Determine the number of racks to be cabled together.
Obtain the total number of InfiniBand cables required for the installation. See Multirack Cabling Tables.
Ensure that each rack has a Sun Datacenter InfiniBand Switch 36 spine switch installed in the bottom rack unit (U1). See "Recovery Appliance X6 and X5 Rack Layout" or "Recovery Appliance X4 Rack Layout", as applicable.
Gather the following tools:
Pen
Note pad
Flashlight
Labels for cables and rack
Cable ties (Oracle recommends hook and loop straps)
See Also:
"Parts for InfiniBand Switches" for information about cable labels
Review the safety guidelines. See Zero Data Loss Recovery Appliance Safety and Compliance Guide.
You can extend Recovery Appliance by adding it to another Recovery Appliance rack. The procedure consists of cabling and configuring the racks together. You can cable the racks with no downtime.
Note the following:
There is some performance degradation while cabling the racks together. This degradation results from reduced network bandwidth, and retransmitting data because of packet loss when a cable is unplugged.
The environment is not a high-availability environment because one leaf switch must be off. All traffic goes through the remaining leaf switch.
Only the existing rack is operational, and the new rack is powered down.
The software running on the systems must not have problems related to InfiniBand restarts.
The new racks must be configured with the appropriate IP addresses to be migrated into the expanded system before cabling. Duplicate IP addresses are not allowed.
The existing spine switch is set to priority 10 during the cabling procedure. This setting gives the spine switch a higher priority than any other switch in the fabric. The spine switch is the first to take the Subnet Manager Master role whenever a new Subnet Manager Master is set during the cabling procedure.
A spine switch is required in each rack when configuring multiple racks on the same InfiniBand fabric. A spine switch can be factory installed before shipment, or it can be installed on site. Recovery Appliance uses a third Sun Datacenter InfiniBand Switch 36 as the spine switch.
Note:
This procedure is specific to Recovery Appliance. It differs from the steps in the Sun Datacenter InfiniBand Switch 36 manual.
Sun Datacenter InfiniBand Switch 36 switch accepts InfiniBand cables from floor or underfloor delivery. Floor and underfloor delivery limits the tension in the InfiniBand cable to the weight of the cable for the rack height of the switch.
Note:
Overhead cabling details are not included in this guide. For details on overhead cabling, contact your Oracle service engineer.
This section describes how to cable together two adjacent racks. The existing rack is R1, and the new rack is R2.
To cable two racks together:
Set the priority of the current, active Subnet Manager Master to 10
on the spine switch:
Log in to any InfiniBand switch on the active system.
Use the getmaster
command to verify that the Subnet Manager Master is running on the spine switch. If it is not, then follow the procedure in "Setting a Spine Switch as the Subnet Manager Master".
Log in to the spine switch.
Use the disablesm
command to stop Subnet Manager.
Use the setsmpriority 10
command to set the priority to 10.
Use the enablesm
command to restart the Subnet Manager.
Repeat step 1.b to ensure the Subnet Manager Master is running on the spine switch.
Ensure the new rack is near the existing rack. The InfiniBand cables must be able to reach the servers in each rack.
Completely shut down the new rack (R2).
Cable the leaf switch in the new rack according to Table E-4.
Power off leaf switch R1 IB2. This action causes all compute servers and storage servers to fail over their InfiniBand traffic to R1 IB3.
Disconnect all interswitch links between R1IB2 and R1 IB3.
Cable leaf switch R1 IB2 according to Table E-3.
Power on leaf switch R1 IB2.
Wait three minutes for R1 IB2 to become completely operational.
To check the switch, log in to it and run the ibswitches
command. The output should show three switches, R1 IB1, R1 IB2, and R1 IB3.
Power off leaf switch R1 IB3. This action causes all compute servers and storage servers to fail over their InfiniBand traffic to R1 IB2.
Cable leaf switch R1 IB3 according to Table E-3.
Power on leaf switch R1 IB3.
Wait three minutes for R1 IB3 to become completely operational.
To check the switch, log in to it and run the ibswitches
command. The output should show three switches, R1 IB1, R1 IB2, and R1 IB3.
Power on all InfiniBand switches in R2.
Wait for three minutes for the switches to become completely operational.
To check the switch, log in to it and run the ibswitches
command. The output should show six switches, R1 IB1, R1 IB2, R1 IB3, R2 IB1, R2 IB2, and R2 IB3.
Run the getmaster
command from any switch to ensure that the Subnet Manager Master is running on R1 IB1.
Power on all servers in R2.
Log in to spine switch R1 IB1, and lower its priority to 8:
Use the disablesm
command to stop the Subnet Manager.
Use the setsmpriority 8
command to set the priority to 8.
Use the enablesm
command to restart the Subnet Manager.
Ensure that the Subnet Manager Master is running on one of the spine switches.
After cabling the racks together, configure them as described in Configuring a Recovery Appliance Rack.
This section describes how to cable several adjacent racks together. The existing racks are R1, R2,... Rn, the new rack is Rn+1, and the Subnet Manager Master is running on R1 IB1.
To cable several racks together:
Set the priority of the current, active Subnet Manager Master to 10
on the spine switch:
Log in to any InfiniBand switch on the active system.
Use the getmaster
command to verify that the Subnet Manager Master is running on the spine switch. If it is not, then follow the procedure in "Setting a Spine Switch as the Subnet Manager Master".
Log in to the spine switch.
Use the disablesm
command to stop the Subnet Manager.
Use the setsmpriority 10
command to set the priority to 10.
Use the enablesm
command to restart the Subnet Manager.
Repeat step 1.b to ensure that the Subnet Manager Master is running on the spine switch.
Ensure that the new rack is near the existing rack. The InfiniBand cables must be able to reach the servers in each rack.
Completely shut down the new rack (Rn+1).
Cable the leaf switch in the new rack according to the appropriate table in Multirack Cabling Tables. For example, if rack Rn+1 is R4, then use Table E-11.
Complete the following procedure for each of the original racks. In these steps, Rx represents a rack number from R1 to Rn.
Power off leaf switch Rx IB2. This action causes all compute servers and storage servers to fail over their InfiniBand traffic to Rx IB3.
Cable leaf switch Rx IB2 according to Multirack Cabling Tables.
Power on leaf switch Rx IB2.
Wait three minutes for Rx IB2 to become completely operational.
To check the switch, log in to it and run the ibswitches
command. The output should show n*3 switches for IB1, IB2, and IB3 in racks R1, R2,... Rn.
Power off leaf switch Rx IB3. This action causes all compute servers and storage servers to fail over their InfiniBand traffic to Rx IB2.
Cable leaf switch Rx IB3 according to Multirack Cabling Tables.
Power on leaf switch Rx IB3.
Wait three minutes for Rx IB3 to become completely operational.
To check the switch, log in to the switch and run the ibswitches
command. The output should show n*3 switches for IB1, IB2, and IB3 in racks R1, R2,... Rn.
Repeat these steps for each rack until all of them are recabled according to Multirack Cabling Tables.
Power on all InfiniBand switches in Rn+1.
Wait three minutes for the switches to become completely operational.
To check the switch, log in to it and run the ibswitches
command. The output should show (n+1)*3 switches for IB1, IB2, and IB3 in racks R1, R2,... Rn+1.
Run the getmaster
command from any switch to ensure that the Subnet Manager Master is running on R1 IB1.
Power on all servers in Rn+1.
Log in to spine switch R1 IB1, and lower its priority to 8:
Use the disablesm
command to stop the Subnet Manager.
Use the setsmpriority 8
command to set the priority to 8.
Use the enablesm
command to restart the Subnet Manager.
To ensure that the Subnet Manager Master is running on one of the spine switches, use the getmaster
command from any switch.
To ensure that the Subnet Manager is running on every spine switch, use the following command from any switch:
ibdiagnet -r
The Summary Fabric SM-state-priority section of the output should show that the spine switches are running.
If a spine switch is not running, then log in to the switch and use the enablesm
command to enable the Subnet Manager.
If there are now four or more racks, then log in to the leaf switches in each rack and use the disablesm
command to disable the Subnet Manager
In addition to the hardware installation of the new racks, you must do the following procedures:
Generate the installation files. See Using the Deployment Assistant.
Configure the rack components. See Configuring a Recovery Appliance Rack.
Install the Recovery Appliance software. See Installing the Recovery Appliance Software.