Updating the system automatically using the CLI

You can use the command-line interface (CLI) to install software updates.

1. You must download, install, and run the latest version of the test utility to verify that no issues exist with the current system.
   Important: After you install and run the test utility by using either the management GUI or the CLI, you must return to step 2 on this page.

   You can download and install the most current version of this tool at the following website. The link provides instructions for using either the management GUI or the CLI to install and run the test utility.

   http://www.ibm.com/support/docview.wss?uid=ssg1S4000585

2. Download the latest code from the www.ibm.com/support site.
   • If you want to write the code to a CD, you must download the CD image.
   • If you do not want to write the code to a CD, you must download the installation image.
3. Use PuTTY scp (pscp) to copy the update files to the node.
4. Ensure that the update file was successfully copied.
   Before you begin the update, you must be aware of the following situations:

   • The installation process fails under the following conditions:
     • If the code that is installed on the remote system is not compatible with the new code or if an intersystem communication error does not allow the system to check that the code is compatible.
     • If any node in the system has a hardware type that is not supported by the new code.
     • If the system determines that one or more volumes in the system would be taken offline by rebooting the nodes as part of the update process. You can find details about which volumes would be affected by using the lsdependentvdisks command. If you are prepared to lose access to data during the update, you can use the force flag to override this restriction.
   • The update is distributed to all the nodes in the system by using internal connections between the nodes.
   • Nodes are updated one at a time.
   • Nodes run the new code concurrently with normal system activity.
   • While the node is updated, it does not participate in I/O activity in the I/O group. As a result, all I/O activity for the volumes in the I/O group is directed to the other node in the I/O group by the host multipathing software.
   • There is a thirty-minute delay between node updates. The delay allows time for the host multipathing software to rediscover paths to the nodes that are updated. There is no loss of access when another node in the I/O group is updated.
   • The update is not committed until all nodes in the system are successfully updated to the new code level. If all nodes are successfully restarted with the new code level, the new level is committed. When the new level is committed, the system vital product data (VPD) is updated to reflect the new code level.
   • Wait until all member nodes are updated and the update is committed before you invoke the new functions of the updated code.
   • Because the update process takes some time, the installation command completes as soon as the code level is verified by the system. To determine when the update is completed, you must either display the code level in the system VPD or look for the Software update complete event in the error/event log. If any node fails to restart with the new code level or fails at any other time during the process, the code level is backed off.
   • During an update, the version number of each node is updated when the code is installed and the node is restarted. The system code version number is updated when the new code level is committed.
   • When the update starts, an entry is made in the error or event log and another entry is made when the update completes or fails.
5. Issue this CLI command to start the update process:

   applysoftware -file software_update_file

   Where software_update_file is the name of the code update file in the directory you copied the file to in step 3.

   If the system identifies any volumes that would go offline as a result of rebooting the nodes as part of the system update, the code update does not start. An optional force parameter can be used to indicate that the update continues regardless of the problem identified. If you use the force parameter, you are prompted to confirm that you want to continue. The behavior of the force parameter changes, and it is no longer required when you apply an update to a system with errors in the event log.
6. If you are updating from a release before version 7.4.0, issue the following CLI command to check the status of the code update process:

   svcinfo lssoftwareupgradestatus

   This command displays inactive when the update is complete.
   Note: If a status of stalled_non_redundant is displayed, proceeding with the remaining set of node updates might result in offline volumes. Contact a service representative to complete the update.
7. If you are updating from version 7.4.0 or later, issue the following CLI command to check the status of the code update process:

   lsupdate

   This command displays success when the update is complete. If you have hot-spare nodes that are configured on your system, the hot-spare node assumes I/O operations from each node as it is updated.
   Note: If a status of stalled_non_redundant is displayed, proceeding with the remaining set of node updates might result in offline volumes. Contact a service representative to complete the update.
8. If you updated from a release before version 7.4.0, you receive the status message system_completion_required. To complete the update process, issue the command applysoftware -complete. After that command is run, you can run lsupdate to see the progress of the update completion.
9. To verify that the update successfully completed, issue the lsnodevpd CLI command for each node that is in the system.

   The code version field displays the new code level.

   Important: If you update your system software to version 8.1.1 or later from a version earlier than 8.1.0, on a system where you have already installed more than 64 GB of RAM, all nodes return from the update with an error code of 841. Versions 8.1.0 and later allocate memory in a different way than previous versions, so the RAM must be "accepted" again. To resolve the error, complete the following steps:

   1. On a single node, run the svctask chnodehw command. Do not run the command on more than one node at a time.
   2. Wait for the node to restart and return without the error.
   3. Wait an additional 30 minutes for multipath drives to recover on the host.
   4. Repeat this process for each node individually until you clear the error on all nodes.