SPARCServer 1000 - Solaris 2.3 -> 2.4 Upgrade??

SPARCServer 1000 - Solaris 2.3 -> 2.4 Upgrade??

Post by Alexis Oosterho » Fri, 03 Feb 1995 14:12:05



Was just wondering if anyone out there had upgraded a SS1000 from
2.3 to 2.4 ?  Advice? Problems? Extra Patches?  Any comments would
be greatly appreciated.

Alexis

-

Systems Administrator        
WNI Science & Engineering and Weathernews Pty Ltd
31 Bishop St, Jolimont W.A 6014 AUSTRALIA  ph:(61 9) 387-7955 fax:387-6686

 
 
 

SPARCServer 1000 - Solaris 2.3 -> 2.4 Upgrade??

Post by Don Turrenti » Sun, 05 Feb 1995 02:07:16


: Was just wondering if anyone out there had upgraded a SS1000 from
: 2.3 to 2.4 ?  Advice? Problems? Extra Patches?  Any comments would
: be greatly appreciated.

: Alexis

: -

: Systems Administrator        
: WNI Science & Engineering and Weathernews Pty Ltd
: 31 Bishop St, Jolimont W.A 6014 AUSTRALIA  ph:(61 9) 387-7955 fax:387-6686

I tried and failed. When I chose the upgrade option it told me that there
were no local disks to upgrade. The only thing that I can think of that
could have caused this (besides a bad install program!) is that my two
internal disks are c0t0d0 and c0t1d0 with the boot disk be c0t0d0. I believe
that Solaris expects c0t3d0 to be the first (boot) disk, This is how my
system arrived from the factory, so if its wrong its the fault of SMCC.

Anyhow I went ahead and did an initial install and when everything was loaded
I couldn't bring up my SparcStorage Array because it required version 2.0 of
the SparcStorage Array software which I didn't recieve until 4 weeks later.
Needless to say I had to restore back to 2.3 :(

--
Don Turrentine, Information Systems Specialist
UAB - Cardiac Rhythm Management Laboratory


 
 
 

SPARCServer 1000 - Solaris 2.3 -> 2.4 Upgrade??

Post by Casper H.S. D » Sun, 05 Feb 1995 05:40:31



>I tried and failed. When I chose the upgrade option it told me that there
>were no local disks to upgrade. The only thing that I can think of that
>could have caused this (besides a bad install program!) is that my two
>internal disks are c0t0d0 and c0t1d0 with the boot disk be c0t0d0. I believe
>that Solaris expects c0t3d0 to be the first (boot) disk, This is how my
>system arrived from the factory, so if its wrong its the fault of SMCC.

Well, I have reasons to believe otherwise.  When we netinstalled a new
server (with 2 F/W/D SCSI controllers) we made a mistake in setting the
boot disk in the PROM.  Auto-install installed on a different disk from the
one we intended to install one.  So we modified the profile and made explicit
which disks where to get which mountpoint in the profile.  This time
auto-install completed, but after installing the packages it bailed
out with a message to the effect: "Your boot disk isn't the one
I just installed Solaris on." followed by a root prompt.
So it does know which disk to use.  It actually looks at all
disks for bootable slice 0 (I think that's the prerequisite)

There are several known cases for upgrade failures:

        - you have filesystems > 2GB (there's a workaround,
          even if you can't remove the filesystems from /etc/vfstab
          because they need to be upgraded.)
        - if you have a lot of installed patches the installtool
          may dump core (this is what happened to us)
        - /var/sadm was moved to a different disk and a symbolic
          link was installed but it wasn't a relative symlink
        - /var/sadm/softinfo is damaged.

Quote:>Anyhow I went ahead and did an initial install and when everything was loaded
>I couldn't bring up my SparcStorage Array because it required version 2.0 of
>the SparcStorage Array software which I didn't recieve until 4 weeks later.
>Needless to say I had to restore back to 2.3 :(

I can see your problem there.

Casper

 
 
 

SPARCServer 1000 - Solaris 2.3 -> 2.4 Upgrade??

Post by Mark Fro » Thu, 09 Feb 1995 12:49:40





>>I tried and failed. When I chose the upgrade option it told me that there
>>were no local disks to upgrade. The only thing that I can think of that
>>could have caused this (besides a bad install program!) is that my two
>>internal disks are c0t0d0 and c0t1d0 with the boot disk be c0t0d0. I believe
>>that Solaris expects c0t3d0 to be the first (boot) disk, This is how my
>>system arrived from the factory, so if its wrong its the fault of SMCC.

>Well, I have reasons to believe otherwise.  When we netinstalled a new
>server (with 2 F/W/D SCSI controllers) we made a mistake in setting the
>boot disk in the PROM.  Auto-install installed on a different disk from the
>one we intended to install one.  So we modified the profile and made explicit
>which disks where to get which mountpoint in the profile.  This time
>auto-install completed, but after installing the packages it bailed
>out with a message to the effect: "Your boot disk isn't the one
>I just installed Solaris on." followed by a root prompt.
>So it does know which disk to use.  It actually looks at all
>disks for bootable slice 0 (I think that's the prerequisite)

>There are several known cases for upgrade failures:

>    - you have filesystems > 2GB (there's a workaround,
>      even if you can't remove the filesystems from /etc/vfstab
>      because they need to be upgraded.)
>    - if you have a lot of installed patches the installtool
>      may dump core (this is what happened to us)
>    - /var/sadm was moved to a different disk and a symbolic
>      link was installed but it wasn't a relative symlink
>    - /var/sadm/softinfo is damaged.

I am trying to upgrade a SS1000 to 2.4 from 2.3 and have found that I
get a panic ("BAD TRAP: memory alignment error" or words to that effect)
whenever I get to the part just before it asks me if I want to upgrade
or do an initial install.

I have yet to get around this.  The Sun support person told me to check
the printenv output from OpenBoot to see if it looks corrupted and if so
that I should clear the env settings.  If this doesn't work I don't know
what to try.

-mark frost