Sol 10 05/08 upgrade - hangs on DVD boot
philkime
quite a while. I've performed a standard upgrade a few times without
problems and it's currently running a fully patched 08/07. However,
when I try to boot off the 05/08 DVD (which boots fine in a VMware
machine), it gets past the grub screen and tries to load the kernel
off the DVD and hangs at the license line. With verbosity, the last
line I get is:
using default device instance data
I've tried the user-acpi-options settings and disabling everything
likely in the BIOS but no luck - it always hangs at the same place.
Any ideas?
John D Groenveld
philkime <Phi...@kime.org.uk> wrote:
>I have an old Dell 500SC which has been running Sol 10 perfectly for
>quite a while. I've performed a standard upgrade a few times without
>problems and it's currently running a fully patched 08/07. However,
>when I try to boot off the 05/08 DVD (which boots fine in a VMware
>machine), it gets past the grub screen and tries to load the kernel
>off the DVD and hangs at the license line. With verbosity, the last
I just hung a Pentium 3 test machine at the same place after applying
127128-11 released yesterday, which I believe is integrated with
S10 5/08.
Added the patch to the ignore list in PCA's pca.conf until its
withdrawn.
John
groe...@acm.org
philkime
> I just hung a Pentium 3 test machine at the same place after applying
> 127128-11 released yesterday, which I believe is integrated with
> S10 5/08.
>
> Added the patch to the ignore list in PCA's pca.conf until its
> withdrawn.
Hmm, in that case I need to make a new miniroot for the DVD without
this patch? I only have one SOL10 machine so I can't do anything with
flars etc. Disks aren't set up for LIveUpgrade either - has to be DVD.
PK
Kostik Belousov
I doubt that this is kernel itself. I have a similar problem with
2xCPU old machine that I reported several days ago to the list.
It hangs at the same place when booted from the DVD.
But, the 127128-11 kernel boots fine from the hard drive.
John D Groenveld
Kostik Belousov <konstanti...@zoral.com.ua> wrote:
>I doubt that this is kernel itself. I have a similar problem with
>2xCPU old machine that I reported several days ago to the list.
>It hangs at the same place when booted from the DVD.
>
>But, the 127128-11 kernel boots fine from the hard drive.
When I patchrm 127128-11 from my dual socket P3 (an ASUS CUV4X-DLS)
it boots.
But 127128-11 is queer: it will only apply in single-user mode.
Hopefully someone with more time will file a service order.
John
groe...@acm.org
usenetper...@gmail.com
> On Apr 29, 8:27 pm, groen...@cse.psu.edu (John D Groenveld) wrote:
> > I just hung a Pentium 3 test machine at the same place after applying
> > 127128-11 released yesterday, which I believe is integrated with
> > S10 5/08.
> > Added the patch to the ignore list in PCA's pca.conf until its
> > withdrawn.
But its the "new" kernel on the DVD right? so....
> Hmm, in that case I need to make a new miniroot for the DVD without
> this patch? I only have one SOL10 machine so I can't do anything with
> flars etc. Disks aren't set up for LIveUpgrade either - has to be DVD.
Have a look at the 14 NOTEs for this kernel patch. It might help.
Otherwise it would seem that you stuck with patching update 4 for now.
The changes in Update 5 may have little relevance to an older machine
anyway??
So no biggie.
John D Groenveld
<usenetper...@gmail.com> wrote:
>Otherwise it would seem that you stuck with patching update 4 for now.
Yes.
>The changes in Update 5 may have little relevance to an older machine
>anyway??
>So no biggie.
No biggie for existing Solaris 10 customers with supported, legacy
Pentium 3 hardware, at least until Sun starts shipping S10 patches which
depend on 127128.
Big biggie for new customers with supported, legacy Pentium 3 hardware.
Any sign of a BugID yet?
BTW no sign in the S10 5/08 release notes that P2 and P3 support
is going away in Solaris Next.
<URL:http://docs.sun.com/app/docs/doc/820-4078/>
But I see that Intel is giving a keynote at JavaONE so may be Anil Gadre
and company has decided to leave it to Intel and AMD to make the case
to moving to Sun's modern x64 systems.
<URL:http://www.sun.com/aboutsun/pr/2008-04/sunflash.20080416.2.xml>
John
groe...@acm.org
usenetper...@gmail.com
> In article <5968f57a-26bd-4be0-9852-b179b8df2...@x19g2000prg.googlegroups.com>,
> <usenetpersonger...@gmail.com> wrote:
> >Otherwise it would seem that you stuck with patching update 4 for now.
> Yes.
> >The changes in Update 5 may have little relevance to an older machine
> >anyway??
> >So no biggie.
> No biggie for existing Solaris 10 customers with supported, legacy
> Pentium 3 hardware, at least until Sun starts shipping S10 patches which
> depend on 127128.
That "depends" on whether the Installation Notes have workarounds
already.. : >
One workaround (boot in 32 bit mode) can be attempted surely.
> Big biggie for new customers with supported, legacy Pentium 3 hardware.
Whats been added to Update 5 that something this ancient would use I
wonder?
> Any sign of a BugID yet?
If the workarounds work then there isnt really a bug.
> BTW no sign in the S10 5/08 release notes that P2 and P3 support
> is going away in Solaris Next.
> <URL:http://docs.sun.com/app/docs/doc/820-4078/>
> But I see that Intel is giving a keynote at JavaONE so may be Anil Gadre
> and company has decided to leave it to Intel and AMD to make the case
> to moving to Sun's modern x64 systems.
> <URL:http://www.sun.com/aboutsun/pr/2008-04/sunflash.20080416.2.xml>
Guys like that dont even live on the same planet as us.. : /
John D Groenveld
<usenetper...@gmail.com> wrote:
>Whats been added to Update 5 that something this ancient would use I
>wonder?
I don't believe its possible for current or potential customers to
download Update 4 for their legacy supported hardware.
>If the workarounds work then there isnt really a bug.
What work-around?
>Guys like that dont even live on the same planet as us.. : /
The Ed Zander business model didn't work out well for Sun.
John
groe...@acm.org
Ian Collins
> On Apr 30, 8:32 am, groen...@cse.psu.edu (John D Groenveld) wrote:
>> In article <5968f57a-26bd-4be0-9852-b179b8df2...@x19g2000prg.googlegroups.com>,
>> <usenetpersonger...@gmail.com> wrote:
>>> Otherwise it would seem that you stuck with patching update 4 for now.
>> Yes.
>>> The changes in Update 5 may have little relevance to an older machine
>>> anyway??
>>> So no biggie.
>> No biggie for existing Solaris 10 customers with supported, legacy
>> Pentium 3 hardware, at least until Sun starts shipping S10 patches which
>> depend on 127128.
>
> That "depends" on whether the Installation Notes have workarounds
> already.. : >
> One workaround (boot in 32 bit mode) can be attempted surely.
>
Maybe it's a ruse to make sure people plan for LU...
--
Ian Collins.
philkime
and it did but it is certainly the culprit and the machine no longer
boots, hanging at the same place as the 05/08 install DVD.
So, I booted into failsafe to remove the patch but it won't remove
since of course it can't start the non-global zone I have on there.
So, I tried the -G flag to patchrm but of course, the patch included
packages with the ALLZONES flag set so it won't work with the -G flag.
Catch-22.
Any ideas?
I did split the root mirror before I applied it but I cannot for the
life of me work out how to get the machine to boot of the other disk
with grub.
Tim Stewart
> In article
> <2384ae16-80e7-4374...@j33g2000pri.googlegroups.com>,
> philkime <Phi...@kime.org.uk> wrote:
>>I have an old Dell 500SC which has been running Sol 10 perfectly for
>>quite a while. I've performed a standard upgrade a few times without
>>problems and it's currently running a fully patched 08/07. However,
>>when I try to boot off the 05/08 DVD (which boots fine in a VMware
>>machine), it gets past the grub screen and tries to load the kernel
>>off the DVD and hangs at the license line. With verbosity, the last
>
> I just hung a Pentium 3 test machine at the same place after
> applying 127128-11 released yesterday, which I believe is integrated
> with S10 5/08.
I've just hung up my Pentium 3 box after patching to 127128-11.
Previously it was U4 with a rather recent patch cluster. I've called
up USA4SUN to see what they have to say about it.
I'll post my results here, when I have something.
--
-TimS
Tim Stewart
Lead UNIX Systems Administrator
Ciena Corporation
Alpharetta, GA, USA
tste...@ciena.com
Tim Stewart
Tim Stewart <tste...@ciena.com> writes:
> groe...@cse.psu.edu (John D Groenveld) writes:
>
>> In article
>> <2384ae16-80e7-4374...@j33g2000pri.googlegroups.com>,
>> philkime <Phi...@kime.org.uk> wrote:
>>>I have an old Dell 500SC which has been running Sol 10 perfectly for
>>>quite a while. I've performed a standard upgrade a few times without
>>>problems and it's currently running a fully patched 08/07. However,
>>>when I try to boot off the 05/08 DVD (which boots fine in a VMware
>>>machine), it gets past the grub screen and tries to load the kernel
>>>off the DVD and hangs at the license line. With verbosity, the last
>>
>> I just hung a Pentium 3 test machine at the same place after
>> applying 127128-11 released yesterday, which I believe is integrated
>> with S10 5/08.
>
> I've just hung up my Pentium 3 box after patching to 127128-11.
> Previously it was U4 with a rather recent patch cluster. I've called
> up USA4SUN to see what they have to say about it.
>
> I'll post my results here, when I have something.
I called USA4SUN and they suggested a workaround that works on my
system. Append `-kd' to the kernel line in GRUB to enter kmdb upon
boot, and type `cmi_no_init/W1', and then `:c' - example listed below:
Loading kmdb...
Welcome to kmdb
kmdb: unable to determine terminal type: assuming `vt100'
[0]> cmi_no_init/W1
cmi_no_init: 0 = 0x1
[0]> :c
SunOS Release 5.10 Version Generic_127128-11 32-bit
Copyright 1983-2008 Sun Microsystems, Inc. All rights reserved.
Use is subject to license terms.
Hostname: bitbox
...
Everything seems to be functioning on this system. The case is still
open with Sun and I'm going to give them as much information as I can
to help them debug.
I'll post more news as I find it.
Andrew Gabriel
Tim Stewart <tste...@ciena.com> writes:
> I called USA4SUN and they suggested a workaround that works on my
> system. Append `-kd' to the kernel line in GRUB to enter kmdb upon
> boot, and type `cmi_no_init/W1', and then `:c' - example listed below:
>
> Loading kmdb...
>
> Welcome to kmdb
> kmdb: unable to determine terminal type: assuming `vt100'
> [0]> cmi_no_init/W1
> cmi_no_init: 0 = 0x1
> [0]> :c
> SunOS Release 5.10 Version Generic_127128-11 32-bit
> Copyright 1983-2008 Sun Microsystems, Inc. All rights reserved.
> Use is subject to license terms.
> Hostname: bitbox
> ...
>
>
> Everything seems to be functioning on this system. The case is still
> open with Sun and I'm going to give them as much information as I can
> to help them debug.
>
> I'll post more news as I find it.
You could make that change permanent by adding following to /etc/system
once you get the system up:
set cmi_no_init=1
--
Andrew Gabriel
[email address is not usable -- followup in the newsgroup]
Tim Stewart
> You could make that change permanent by adding following to
> /etc/system once you get the system up:
>
> set cmi_no_init=1
Great, I'll do that. Thanks!
Andrew Gabriel
Tim Stewart <tste...@ciena.com> writes:
> and...@cucumber.demon.co.uk (Andrew Gabriel) writes:
>
>> You could make that change permanent by adding following to
>> /etc/system once you get the system up:
>>
>> set cmi_no_init=1
>
> Great, I'll do that. Thanks!
Tim, I also emailed you some more questions.
John D Groenveld
Tim Stewart <tste...@ciena.com> wrote:
>> I've just hung up my Pentium 3 box after patching to 127128-11.
>> Previously it was U4 with a rather recent patch cluster. I've called
>> up USA4SUN to see what they have to say about it.
>>
>> I'll post my results here, when I have something.
>
>I called USA4SUN and they suggested a workaround that works on my
>system. Append `-kd' to the kernel line in GRUB to enter kmdb upon
Thank you very much for filing a service order on this bug!
Do you have a BugID?
John
groe...@acm.org
philkime
> You could make that change permanent by adding following to /etc/system
> once you get the system up:
>
> set cmi_no_init=1
I decided to do this properly and set up a LU environment on a
mirrored pair of slices. I then upgraded this to 05/08 and it hung as
expected due to 127128-11. Using kmdb as described, it booted and
after adding the above to /etc/system, it booted without problems.
Many thanks to all concerned. I am not sure what impact disabling the
CPU feature init has but I can't see anything noticeable so far. I
also wanted to point out that LU upgrades to 05/08 require patch
137322-01 (x86) first otherwise the upgrade will report that 600+
packages failed to install because they are in an updated bzip format
which this patch provides. That caught me the first time I did the
upgrade. It is in the release notes for 05/08 but not in a
particularly obvious place.
i5mast
> system. Append `-kd' to the kernel line in GRUB to enter kmdb upon
> boot, and type `cmi_no_init/W1', and then `:c' - example listed below:
>
> Loading kmdb...
>
> Welcome to kmdb
> kmdb: unable to determine terminal type: assuming `vt100'
> [0]>cmi_no_init/W1
> cmi_no_init: 0 = 0x1
> [0]> :c
> SunOS Release 5.10 Version Generic_127128-11 32-bit
> Copyright 1983-2008 Sun Microsystems, Inc. All rights reserved.
> Use is subject to license terms.
> Hostname: bitbox
> ...
Thanks. This fixed the hang on boot issue for me after 127128-11. My
box is Dell Optiplex GX150 - Pentium III.
--
i5mast