Problems with Sol10 patch cluster?

Problems with Sol10 patch cluster?

Post by Stuar » Sun, 26 Feb 2006 19:16:41



I've installed the latest recommended patch cluster for Solaris 10 on a
U60 and observed a number of retun codes =1.

Is there a major problem here?

Stuart

 
 
 

Problems with Sol10 patch cluster?

Post by Stuar » Sun, 26 Feb 2006 19:51:03



> I've installed the latest recommended patch cluster for Solaris 10 on a
> U60 and observed a number of retun codes =1.

> Is there a major problem here?

> Stuart

Figuring the worst that could is that I reinstall the OS, I rebooted.
It seems to be OK.

For now.

Looking at the logs I found a few messages to the effect that the
architecture for a certain package differs from the package installed
on the system....

Now.. I know I didn't install x86 patches on the U60... :-)

Stuart

 
 
 

Problems with Sol10 patch cluster?

Post by Martin Pau » Tue, 28 Feb 2006 20:44:13



> Looking at the logs I found a few messages to the effect that the
> architecture for a certain package differs from the package installed
> on the system....

It's probably patches like 121236-01 that showed this message:

  Architecture for package SUNWcakr from directory SUNWcakr.v in patch
  121236-01 differs from the package installed on the system.

The architecture it talks about is the kernel architecture (arch -k).
The patch is for sun4v (niagara machines ?) only, while your U60 has
sun4u architecture. So it actually means that the patch is for a
packages which is not installed on your machine, and therefore can
safely be ignored.

mp.
--
Systems Administrator | Institute of Scientific Computing | Univ. of Vienna

 
 
 

Problems with Sol10 patch cluster?

Post by Stuar » Wed, 01 Mar 2006 17:00:10




> > Looking at the logs I found a few messages to the effect that the
> > architecture for a certain package differs from the package installed
> > on the system....

> It's probably patches like 121236-01 that showed this message:

>   Architecture for package SUNWcakr from directory SUNWcakr.v in patch
>   121236-01 differs from the package installed on the system.

> The architecture it talks about is the kernel architecture (arch -k).
> The patch is for sun4v (niagara machines ?) only, while your U60 has
> sun4u architecture. So it actually means that the patch is for a
> packages which is not installed on your machine, and therefore can
> safely be ignored.

Thanks Martin.

That makes sense; I guess I was expecting that would yeild a different
return code than "1"

Stuart

 
 
 

Problems with Sol10 patch cluster?

Post by Martin Pau » Wed, 01 Mar 2006 17:15:06



> That makes sense; I guess I was expecting that would yeild a different
> return code than "1"

Indeed. According to patchadd source, return code 1 means "Usage error",
which is kind of strange. In the past, I noticed that there's something
fishy about patchadd's return codes in Solaris 10, but I don't know if this
started with FCS or came in with one of the SunUpdateConnection patches, and
whether it's a bug or intentional.

The patchadd man page documents a new option:

  -t
     Maintains the patchadd return  codes  from  the  Solaris
     release  prior  to Solaris 10. On a system with zones(5)
     installed, a return code of  0  indicates  success.  Any
     other return code indicates failure.

This results in:

  # patchadd -t 121236-01
  Transition old-style patching.

  Checking installed patches...

  Patchadd is terminating.
  # echo $?
  8

return code 8 is "Attempting to patch a package that is not installed",
which makes more sense. On the other hand, patchadd -t doesn't show a
verbose error message on why it terminated.

Maybe someone can fill us in on why this changes happened.

mp.
--
Systems Administrator | Institute of Scientific Computing | Univ. of Vienna