For use with Version 8 Release 8.4.0

Date: 22 August 2016
(C) Copyright International Business Machines Corp., 2016 All rights reserved.

Contents

The information in this Readme contains fix list and other package information about the Hardware Management Console.

• Description
• Package Information
• Known Issues
  
• General fixes
• Security fixes
• Installation

This package represents a service pack image that can be used to update your HMC from HMC V8 R8.4.0 to HMC V8 R8.4.0 Service Pack 2.  You can also reference this package by PTF MH01620 and APAR MB04005.  This image can be installed on top of HMC Version 8 Release 8.4.0 Recovery installation (PTF MH01559) with Mandatory PTF MH01560 installed.

Note: Service packs are cumulative and as such will include all the fixes for the PTFs released up to and including the last service pack(s) for this HMC version.  Please read the individual Readme files for each PTF to see the list of fixes.

• Enhanced the chhmcencr and lshmcencr commands to support user configuration of the encryption ciphers and Message Authentication Code (MAC) algorithms used by the HMC Secure Shell (SSH) interface. 

This package provides the following fixes:

• Changed SRC E212E139 so that it is no longer called home.  This SRC is commonly generated when the HMC is used to install VIOS partitions even when the installations are successful.
  
• Fixed an issue where a managed system may go into an Incomplete state after removing I/O hardware due to stale data in the hardware discovery (IOR) cache causing an unhandled NullPointerException.
  
• Fixed a rare problem that can result in the server having a persistent Incomplete state due to an issue reading the server's hardware discovery (IOR) cache.
  
• Fixed a report of serviceable event E3551081 for a temporary error accessing the Remote HMC menu locally on the HMC to be an informational SRC and not call home.
  
• Fixed a problem where the HMC incorrectly logs the errors "The following operation failed to start within the specified time window" and "A SCHEDULED OPERATION EXPIRED" for a scheduled operation that completed successfully and did not expire.  This problem occurred when the scheduled operation failed to complete, not start, within the specified start time window.   It most commonly impacts long running scheduled operations such as backup critical console data.
  
• Fixed an issue that caused the HMC to repeatedly call home SRC E35A0021 to report that the Solid DB size exceeded the threshold limit of 80% even though the threshold limit had not actually been exceeded.
  
• Improved the error message displayed when attempting to activate an IBM i partition with assigned physical I/O on a managed system that does not support native I/O for IBM i.
  
• Fixed a problem where create system plan on some server models including 9117-MMD may fail with error "The command mksysplan not successful due to unknown errors Cause: null".
  
• Fixed a problem causing some rows in various tables in the Classic GUI to disappear when filtering and sorting functions for a table are used together.
  
• Fixed a problem where, on the local HMC console session, the HMC Management >  Open Restricted Shell Terminal GUI task is not displayed for any HMC user except hscroot.
  
• Corrected a problem where clicking the OK button on the Change Network Settings panel has no affect, no error is displayed.  This problem occurs when: 1. A user started the HMC setup wizard but cancelled without completing it; and 2. The user did not fill in or clear the default gateway settings on the routing tab before attempting to save the settings.  
  
• Fixed a rare timing issue where a server may incorrectly be shown in No Connection state when actually only the secondary FSP is in a no connection state.  Resetting the connection to the primary FSP or rebooting the HMC will resolve this issue.
  
• Fixed an issue that impacts partition profiles that contain SR-IOV logical ports and also contain virtual Ethernet adapters that use non-default vSwitches. After the first activation of the profile, any non-default vSwitches used by virtual Ethernet adapters in the profile will be changed to the default vSwitch.  The user must edit all affected profiles to restore the original vSwitch names.
  
• Added the return code definitions to the updhmc man page to clarify that non-zero return codes may not imply an error occurred.
  
• Enhanced the shutdown procedure on HMC models CR5, CR6, CR7, CR8 and CR9 to notify IMM that a power down is going to occur prior to actually halting and powering off or rebooting the HMC.
  
• Fixed a problem where HMC to HMC communication intermittently fails resulting in serviceable event B3036620 .  Other symptoms include failure to negotiate a primary HMC for problem analysis which can result in failure to report a server serviceable event or calling home the same event twice.  Repeated occurrences of the B3036620 without a HMC reboot can eventually lead to a hang of the HMC where users are unable to login via the GUI or run commands via ssh.
  
• No longer initialize the OS version the HMC shows for VIOS partitions with the underlying OS kernel version information.  This practice sometimes caused the OS version shown for VIOS partitions to be the base AIX version, or to be the VIOS OS version appended with the AIX OS version distribution number.  After this fix, if the HMC is unable to query the OS version information from the VIOS, the OS version shown will be blank.
  
• Fixed an issue with the lshwres command that sometimes caused incorrect or empty topology to be shown for virtual Fibre Channel adapters.
  
• Changed the backup critical console data function to no longer fail when the backup of performance monitoring data fails.  With this fix, a warning is reported if the performance monitoring data is not successfully backed up and the backup function continues.  Prior to this fix, the backup function failed with the wrong error.
  
• Fixed a problem causing the WLP server not to start after the HMC is rebooted, causing the REST API functions to not be available.  This impacts the enhanced GUI login,
  PowerVC, PCM and any other function that utilizes the REST API on the HMC.  This problem only occurs if the user runs the save upgrade data task and subsequently reboots the HMC without actually performing an HMC upgrade.  This fix prevents the problem from occuring again and also repairs HMCs previously impacted.
  
• Updated the expiration date for the vterm applet.  The current certificate expires August 25th 2016.
  
• Fixed an issue where, after successfully applying a concurrent server firmware update,  the HMC Change Licensed Internal Code panel could show an incorrect pending deferred firmware level.  The problem does not impact the GUI view levels task or lslic command.  This issue is only exposed by a rare type of concurrent server firmware update which has never been released in the field but could occur in the future.
  
• Fixed an issue with the Change License Internal Code task where specifying an update of only I/O microcode incorrectly updated the server firmware to the latest level even though it was not selected.  This problem only occurs when "latest concurrent" or "latest, even if disruptive" LIC types are selected.  To work around this problem, select the "specific levels" LIC type.
  
• Fixed an issue where HMC backups to a remote server may fail with rc=26 permission denied when the remote user has write access to the target.  The problem only occurs when a previous backup was done and the remote user does not have the permissions to overwrite an existing RemoteAccessFile.Test file.
  
• Fixed an issue with the Wake-on-LAN feature with HMC model 7042-CR9.  The fix requires the Wake-on-LAN feature to be disabled in the system settings for the Ethernet adapter that will be used to receive the wake on magic packet.
  
• Fixed a problem that causes many of the panels on the Enhanced GUI to show characters from the wrong locale.  This problem occurs when the browser language is different from the HMC language.
  
• Fixed a problem causing the lpar_rmc_comm_ifs value to be reset to "all" after an HMC update.  This problem only occurs after an HMC update.  The lpar_rmc_comm_ifs value is correctly persisted after an HMC upgrade.
  
• Fixed an intermittent locking issue that can cause partition operations such as partition migration operations to hang in "Migrating-Running" state and future migration attempts to return "command failed" due to the lock conflict.
  
• Prevent the call home of SRC E302FA10.  This SRC is generated when the SR-IOV firmware update task is launched on a server that does not support SR-IOV.
  
• Prevent the call home of SRC E302FA06.  This SRC is generated when there is a survery error of available Power firmware.
  

Security fixes

• Fixed OpenSSH vulnerabilities: CVE-2015-6563, CVE-2015-6564, CVE-2016-3115 and CVE-2016-1908.
• Fixed IBM Websphere Application Server (WAS) vulnerability: CVE-2016-2923.
• Fixed NTP vulnerabilities: CVE-2015-7703, CVE-2016-1547, CVE-2016-1548, CVE-2016-1550, and CVE-2016-2518.
• Fixed Apache Tomcat vulnerability: CVE-2016-3092.
  

Special Install Instructions

a. The enhanced GUI is not supported for install.  Installing this PTF using the Enhanced+ interface may hang.

b.  Classic GUI install from USB requires Service Pack 0 fix level MH01588 or later.   To install from USB at earlier levels use CLI updhmc command.  See the updhmc "man" page (e.g. man updhmc) for further information and  examples on using the command.

To use the classic GUI install:
1.    Log in again selecting the Log In option of "Classic".
2.    If already logged in to the HMC using Enhanced GUI, log off the HMC.
3.    Install using the normal installation instructions.

Installation instructions for HMC Version 8 upgrades and corrective service can be found at these locations:

Upgrading or restoring HMC Version 8

Installation methods for HMC Version 8 fixes

Instructions and images for upgrading via a remote network install can be found here: