Patch 118844-28 fails to install on Ultra 20?
We're trying to run the standard updatemanager on our Ultra 20 running Solaris 10 x86. We've installed patches a number of times but we're hitting a wall with the recent kernel patch, 118844-28.
This patch shows up as needing to be applied in both the updatemanager gui as well as when we run smpatch analyze. The problem is that it doesn't seem like it will install even though the tools say that the installation procedure was successful. Specifically: we instruct the gui to install the patch, it goes off and looks like it's installing, it comes back and says that the installation is complete and successful and that we need to reboot the machine, we tell it to reboot the machine and it does. But when the machine comes back up the updatemanager (and smpatch) still says that this patch needs to be applied. And showrev says that we're still back at version 27. So it seems like it's properly applying the patch but then the patch isn't actually applied.
We tried to install the patch by hand with smpatch add in single user mode and we had a similar experience. The only exception being that the patch process complained that we had to have patch 117435-02 installed (which we do).
Has anybody seen this before? Has anybody gotten this patch to install on this hardware?
[1311 byte] By [
karlp] at [2007-11-26 6:05:26]

# 2
Hi.
> We note that you have submitted a Sun support case
> for this issue.
I found problem in 118844-28/prepatch script.
Look for: bootdev=`prtconf -v /devices | sed -n '/boot-device/{n;p;}' | grep pci | cut -f 2 -d \'`
And replace "boot-device" for "bootpath".
As I see in prtconf -v /devices:
....
name='bootpath' type=string items=1
value='/pci@0,0/pci-ide@11,1/ide@0/cmdk@0,0:a'
....
name='bios-boot-device' type=string items=1
value='80'
:-)
# 10
Is it just me or are the (kernel) patches hardly tested by SUN? If I look at the recent history of releases of 118844, I noticed some versions with major problems which renders your server useless. These problems are fixed but introduce other major problems.
Since years I use SunSolve and always trusted the patches basically blindly, but recently, I hardly dare to install a patch.
Anyone else the same feeling?
Marcel
# 12
Yes, I agree with that fix- I just did it a little opposite:
# eeprom boot-device=/pci@0,0/pci-ide@11,1/ide@0/cmdk@0,0:a
which added an extra alias (permanently) to my 'BIOS' settings-
# eeprom | grep boot
bootpath=/pci@0,0/pci-ide@7/ide@0/cmdk@0,0:a
boot-device=/pci@0,0/pci-ide@7/ide@0/cmdk@0,0:a
#
I re-ran the 10 Rec. Cluster and it not only installed the kernel patch, but, a handful of other patches in the bundle too.
--David