Shutting down and powering up ESS

The ESS components and frame may need to be powered off in cases such as data center maintenance, relocation, or emergencies. Use the following information to shut down and power up ESS.

Shutting down ESS

  1. Verify that the file systems are not needed by users during the time the system will be unavailable.
  2. If you are using a remote cluster to mount the ESS file system, unmount the file system by issuing the mmumount command from the remote client nodes.
  3. Shut down the nodes using the mmshutdown -N command. For example:
    mmshutdown -N ems1,gssio1,gssio2
  4. If other nodes are attached and ESS nodes are the only quorum and manager nodes, it is recommended that you use the mmshutdown -a command to shut down the entire cluster.
  5. Verify that IBM Spectrum Scale is shut down on the I/O nodes by issuing the mmgetstate -a command.
  6. Power off the EMS and I/O nodes by issuing the mmshutdown -h now command on each individual node.
    If you are using the Big Endian (BE) platform:
    1. The EMC LPAR, I/O node1 LPAR, and I/O node 2 LPAR will be shut down after you issue the shutdown -h now.
    2. Use the HMC to shut down the physical servers.
    3. Verify that the power light on the front of the frame is blinking after the LPARs are shut down.

    If you are using the Big Endian (BE) platform and the HMC resides within this frame:

    1. Power off the HMC. If the HMC controls servers that are outside of this frame, plan appropriately before shutting down.

    If you are using the Little Endian (LE) platform:

    1. The EMC LPAR, I/O node1 LPAR, and I/O node 2 LPAR will be completely shut down after you issue the shutdown -h now command.
    2. Verify that the power light on the front of the frame is blinking.
  7. Power off all storage by flipping the power switches to off.
  8. Before shutting off power to the frame, verify there are no components within the frame that are relied on by external infrastructure such as IB or Ethernet switches. If any of these exist and hardware outside of the frame needs access, plan appropriately before shutting off power to the frame.

Powering up ESS

  1. Verify that power is connected to the frame.
  2. Turn on all PDUs within the ESS frame.
  3. Power on the components in the following order.
    If you are using the Big Endian (BE) platform:
    1. Power on the HMC.
    2. Power on the storage drawers by flipping the power switches on each storage module to on.
    3. Power on the EMS node, I/O node 1 and I/0 node 2.
    4. Wait for the HMC to come online and log in.
    5. Wait for the EMS node, I/O node 1 and I/0 node 2 to be accessible to the HMC.
    6. Once the EMS sees that node, I/O node 1 and I/0 node 2 are powered on, move to the LPAR view for each and power on the associated LPARs:

      EMS LPAR

      1/O node 1 LPAR

      I/O node 2 LPAR

    7. Once all LPARs are powered on, ssh to the EMS node and verify that IBM Spectrum Scale has come online by issuing mmgetstate -N ems1,gssio1,gssio2. If IBM Spectrum Scale does not automatically start, start it manually by issuing mmstartup -N ems1,gssio1,gssio2.
    8. Issue the gnrhealthcheck and the mmhealth cluster show commands, and check the GUI event logs.
    If you are using the Little Endian (LE) platform:
    1. Power on the storage drawers by flipping the power switches on each storage module to on.
    2. Power on the EMS node, I/O node 1 and I/0 node 2.
    3. Once all LPARs are powered on, ssh to the EMS node and verify that IBM Spectrum Scale has come online by issuing mmgetstate -N ems1,gssio1,gssio2. If IBM Spectrum Scale does not automatically start, start it manually by issuing mmstartup -N ems1,gssio1,gssio2.
    4. Issue the gnrhealthcheck and the mmhealth cluster show commands, and check the GUI event logs.