HMC Update to 7.7.9.0 SP1

Like before with HMC Update to 7.7.5.0, HMC Update to 7.7.7.0 SP1 and HMC Update to 7.7.8.0, the recent HMC update to v7.7.9.0 was again not installable directly from the ISO images via the HMC GUI. Along with the HMC network installation images which are now mentioned in the release notes, there is now also an official documentation of the update procedure using the HMC network installation images. It's called “HMC network installation” and provides a more remote admin friendly way of performing the update. Since it's only a slightly shortened version of the procedure i already tested and used in HMC Update to 7.7.7.0 SP1, i decided to stick with my own procedure.

Also a turn for the better is, that now the release notes as well as FixCentral clearly point out the dependencies between the fixpacks that are supposed to go on top of the update release and the order they are supposed to be applied in. In case of MH01405 (aka V7R7.9.0.0) this is currently only MH01428 (aka V7R7.9.0.1 or v7.7.9.0 SP1.

As always, be sure to study the release notes thoroughly before an update attempt. Depending on your environment and HMC hardware there might be a road block in there. Special attention deserves item 2 in the “upgrade notes” section of the release notes of MH01405. Due to a bug in certain Power Systems firmware levels, there is a dependency between the usable HMC version and the firmware version on the managed system. Be sure to check if your managed systems firmware levels fall in the affected version ranges. For me this meant first upgrading certain managed systems to firmware levels not affected by the above bug. Then upgrading the HMCs to v7.7.9.0 SP1. And finally upgrading certain other managed systems to firmware levels depending on HMC version v7.7.9.0 or later. The latter dependency was - besides the fixed security issues - the main reason to upgrade the HMC in the first place.

For reference purposes, here are some example screen shots from a KVM session to the HMC during a update to MH01405:

HMC network based upgrade to v7.7.9.0 - 1

HMC network based upgrade to v7.7.9.0 - 2

HMC network based upgrade to v7.7.9.0 - 3

HMC network based upgrade to v7.7.9.0 - 4

HMC network based upgrade to v7.7.9.0 - 5

HMC network based upgrade to v7.7.9.0 - 6

HMC network based upgrade to v7.7.9.0 - 7

HMC network based upgrade to v7.7.9.0 - 8

HMC network based upgrade to v7.7.9.0 - 9

HMC network based upgrade to v7.7.9.0 - 10

HMC network based upgrade to v7.7.9.0 - 11

After the upgrade to v7.7.9.0 (MH01405) was complete and the HMC had rebooted, the following errors showed up:

Error after HMC upgrade to v7.7.9.0 - 1

Error after HMC upgrade to v7.7.9.0 - 2

Error after HMC upgrade to v7.7.9.0 - 3

There are not a lot of resources on this particular error, but it seems to occationally have occurred on earlier HMC upgrades as well. Up to now there seems to be no negative effect resulting from this error.

After the upgrade to V7R7.9.0.0 (MH01405) is complete, you can apply the service pack V7R7.9.0.1 (MH01428) in the usual way via the HMC GUI. For me the service pack showed the following output during the update process:

The MH01428 update still shows error messages with regard to symlink creation appearing during the update process, which can be savely ignored. The error message “cp: cannot stat `.dev': No such file or directory” is also becoming an old friend and originates from the shell script /images/installImages inside the MH01428 installation ISO image, where non-existing files are attempted to be copied.

The strange output lines starting with “D:”, “+” and “++” characters seemed awfully familiar, just like the output of a “rpm -vv” command. From tracing down the call order of /images/installImages inside the MH01428 installation ISO image:

/images/installImages
448
449 if [ -f $image/installK2Payloads.sh ]
450 then
451     echo "Installing HMC REST Web Services ..."
452     $image/./installK2Payloads.sh $image
453 fi
454

to:

/images/installK2Payloads.sh
  4
  5 LogFile=/tmp/K2Install.log
  6 IndexFile=K2Payloads.txt
  7
[...]
 73     while read line
 74     do
 75     rpmToInstall=`echo $line | cut -d '/' -f 4`
 76     rpmToUpdate=`echo $line | cut -d '/' -f 4 | cut -d '-' -f 1`
 77 
 78     #   pmc.sem.war needs to be uninstalled and then installed
 79     if [ "$rpmToUpdate" == "pmc.sem.war" ]; then
 80         /bin/rpm -evv $rpmToUpdate --nodeps --allmatches
 81     fi
 82 
 83     /bin/rpm -q $rpmToUpdate
 84     if [ $? -eq 0 ]; then
 85         /bin/rpm -vvh -U $rpmToInstall >> $LogFile 2>&1
 86     else
 87         /bin/rpm -vv -i $rpmToInstall >> $LogFile 2>&1
 88     fi
 89     done < $IndexFile

it seems the dubious output is due to a missing output redirection at the “rpm -evv pmc.sem.war” command in line 80 in the above code snippet of “/images/installK2Payloads.sh”. Since the uninstall of pmc.sem.war is mentioned in the comment in line 78 and the uninstall process apparently finished successfully it can be savely ignored. Although from the above output and without privileged access to the HMCs OS its hard to verify, whether the RPM “pmc.sem.war-7.7.9.1-20140905T1656.i386.rpm” was subsequently successfully re-installed. It can indirectly be verified by looking at the log entries written to the file “/tmp/K2Install.log” in line 87 of the above code snippet. The contents of the log file “/var/hsc/log/EZBuild.rpm.log” written from within the RPM install-/erase-time scriptlets could also be interesting in order to verify everything was re-installed properly.

Aside from the error and dubious messages mentioned above, we're currently not experiencing any issues with the new HMC version v7.7.9.0 SP1 (V7R7.9.0.1).