Prev: Host power on is disabled due to:SCC is not valid or is not present
Next: Incoherent ssh daemon in Solaris 10!
From: Stephan Windmüller on 22 Dec 2008 10:44
Hello!

About two weeks ago I installed the current Solaris Os on two v240 to
form a cluster. For the first install I used ZFS, but that was
incompatible with the cluster software. So I reinstalled the OS, chose
"UFS" as the filesystem and mirrored all partitions onto the second
drive in each machine with the Solaris Volume Manager.

Today I wanted to patch these machines and rebooted one of them into
single user mode. Instead of booting correctly, I got this message:

| WARNING: The following files in / differ from the boot archive:
|
| changed /kernel/drv/did.conf
|
| The recommended action is to reboot to the failsafe archive to correct
| the above inconsistency. To accomplish this, on a GRUB-based platform,
| reboot and select the "Solaris failsafe" option from the boot menu.
| On an OBP-based platform, reboot then type "boot -F failsafe". Then
| follow the prompts to update the boot archive. Alternately, to continue
| booting at your own risk, you may clear the service by running:
| "svcadm clear system/boot-archive"

So I booted the machine into failsafe mode and used the instructions
from http://docs.sun.com/app/docs/doc/817-1985/gglbw?a=view to update
the boot-archive. During this procedure I was confused that my first
disk was listed as c0t0d0s0 and not c1t0d0s0. Also the output of
metastat did not work and the system asked me if I wanted to mount a
zpool which it found during startup. But there should not be any more
zpool now, all partitions are UFS.

The reboot did not work with c0t0d0s0 in the vfstab, so I rebooted once
again into failsafe. From this point, the whole system is corrupted. I
am unable to boot anymore. The OpenBoot prompt tells me this:

| Rebooting with command: boot
| Boot device: disk File and args:
| ERROR: Last Trap: Fast Data Access MMU Miss
|
| {1} ok boot
| ERROR: /pci(a)1c,600000: Last Trap: Fast Data Access MMU Miss
|
| {1} ok

I am unsure how to proceed from this point. The second machine tells me
exactly the same error about "/kernel/drv/did.conf", so I do not think
that in both servers a hard drive failed simultaniously.

Regards
Stephan
From: Richard B. Gilbert on 22 Dec 2008 11:36
Stephan Windm�ller wrote:
> Hello!
>
> About two weeks ago I installed the current Solaris Os on two v240 to
> form a cluster. For the first install I used ZFS, but that was
> incompatible with the cluster software. So I reinstalled the OS, chose
> "UFS" as the filesystem and mirrored all partitions onto the second
> drive in each machine with the Solaris Volume Manager.
>
> Today I wanted to patch these machines and rebooted one of them into
> single user mode. Instead of booting correctly, I got this message:
>
> | WARNING: The following files in / differ from the boot archive:
> |
> | changed /kernel/drv/did.conf
> |
> | The recommended action is to reboot to the failsafe archive to correct
> | the above inconsistency. To accomplish this, on a GRUB-based platform,
> | reboot and select the "Solaris failsafe" option from the boot menu.
> | On an OBP-based platform, reboot then type "boot -F failsafe". Then
> | follow the prompts to update the boot archive. Alternately, to continue
> | booting at your own risk, you may clear the service by running:
> | "svcadm clear system/boot-archive"
>
> So I booted the machine into failsafe mode and used the instructions
> from http://docs.sun.com/app/docs/doc/817-1985/gglbw?a=view to update
> the boot-archive. During this procedure I was confused that my first
> disk was listed as c0t0d0s0 and not c1t0d0s0. Also the output of
> metastat did not work and the system asked me if I wanted to mount a
> zpool which it found during startup. But there should not be any more
> zpool now, all partitions are UFS.
>
> The reboot did not work with c0t0d0s0 in the vfstab, so I rebooted once
> again into failsafe. From this point, the whole system is corrupted. I
> am unable to boot anymore. The OpenBoot prompt tells me this:
>
> | Rebooting with command: boot
> | Boot device: disk File and args:
> | ERROR: Last Trap: Fast Data Access MMU Miss
> |
> | {1} ok boot
> | ERROR: /pci(a)1c,600000: Last Trap: Fast Data Access MMU Miss
> |
> | {1} ok
>
> I am unsure how to proceed from this point. The second machine tells me
> exactly the same error about "/kernel/drv/did.conf", so I do not think
> that in both servers a hard drive failed simultaniously.
>
> Regards
> Stephan


The error message you get on attempting to boot strongly suggests a
hardware problem. Call Sun and get it fixed!

I hope you have a backup of your disk from before you started tinkering!
You may need that after you get your MMU problem straightened out.
From: ndjilinski on 23 Dec 2008 02:25
On Dec 23, 2:44 am, Stephan Windmüller <usenet-reply-2...(a)white-
hawk.de> wrote:
> Hello!
>
> About two weeks ago I installed the current Solaris Os on two v240 to
> form a cluster. For the first install I used ZFS, but that was
> incompatible with the cluster software. So I reinstalled the OS, chose
> "UFS" as the filesystem and mirrored all partitions onto the second
> drive in each machine with the Solaris Volume Manager.
>
> Today I wanted to patch these machines and rebooted one of them into
> single user mode. Instead of booting correctly, I got this message:
>
> | WARNING: The following files in / differ from the boot archive:
> |  
> |      changed /kernel/drv/did.conf
> |      
> | The recommended action is to reboot to the failsafe archive to correct
> | the above inconsistency. To accomplish this, on a GRUB-based platform,
> | reboot and select the "Solaris failsafe" option from the boot menu.
> | On an OBP-based platform, reboot then type "boot -F failsafe". Then
> | follow the prompts to update the boot archive. Alternately, to continue
> | booting at your own risk, you may clear the service by running:
> | "svcadm clear system/boot-archive"
>
> So I booted the machine into failsafe mode and used the instructions
> fromhttp://docs.sun.com/app/docs/doc/817-1985/gglbw?a=viewto update
> the boot-archive. During this procedure I was confused that my first
> disk was listed as c0t0d0s0 and not c1t0d0s0. Also the output of
> metastat did not work and the system asked me if I wanted to mount a
> zpool which it found during startup. But there should not be any more
> zpool now, all partitions are UFS.
>
> The reboot did not work with c0t0d0s0 in the vfstab, so I rebooted once
> again into failsafe. From this point, the whole system is corrupted. I
> am unable to boot anymore. The OpenBoot prompt tells me this:
>
> | Rebooting with command: boot
> | Boot device: disk  File and args:
> | ERROR: Last Trap: Fast Data Access MMU Miss
> |
> | {1} ok boot
> | ERROR: /pci(a)1c,600000: Last Trap: Fast Data Access MMU Miss
> |
> | {1} ok
>
> I am unsure how to proceed from this point. The second machine tells me
> exactly the same error about "/kernel/drv/did.conf", so I do not think
> that in both servers a hard drive failed simultaniously.
>
> Regards
>  Stephan

Hi,

not sure what you have done on those servers exactly - but the error
you got (boot_archive and GRUB) is related to x86 architecture. but
you have got v240 (SPARC).

now, if I were in your shoes I would boot off OS CD or net into single-
user mode and:
* restore bootblk on the internal disks
* mount your root FS from the internal disks and look into the
following:
/etc/system
/etc/vfstab

if you get the same error (MMU Miss) when booting off CD or net - open
the call in Sun support.

Cheers,
Nikolai
From: ndjilinski on 23 Dec 2008 20:28
On Dec 23, 6:25 pm, ndjilinski <ndjilin...(a)gmail.com> wrote:
> On Dec 23, 2:44 am, Stephan Windmüller <usenet-reply-2...(a)white-
>
>
>
> hawk.de> wrote:
> > Hello!
>
> > About two weeks ago I installed the current Solaris Os on two v240 to
> > form a cluster. For the first install I used ZFS, but that was
> > incompatible with the cluster software. So I reinstalled the OS, chose
> > "UFS" as the filesystem and mirrored all partitions onto the second
> > drive in each machine with the Solaris Volume Manager.
>
> > Today I wanted to patch these machines and rebooted one of them into
> > single user mode. Instead of booting correctly, I got this message:
>
> > | WARNING: The following files in / differ from the boot archive:
> > |  
> > |      changed /kernel/drv/did.conf
> > |      
> > | The recommended action is to reboot to the failsafe archive to correct
> > | the above inconsistency. To accomplish this, on a GRUB-based platform,
> > | reboot and select the "Solaris failsafe" option from the boot menu.
> > | On an OBP-based platform, reboot then type "boot -F failsafe". Then
> > | follow the prompts to update the boot archive. Alternately, to continue
> > | booting at your own risk, you may clear the service by running:
> > | "svcadm clear system/boot-archive"
>
> > So I booted the machine into failsafe mode and used the instructions
> > fromhttp://docs.sun.com/app/docs/doc/817-1985/gglbw?a=viewtoupdate
> > the boot-archive. During this procedure I was confused that my first
> > disk was listed as c0t0d0s0 and not c1t0d0s0. Also the output of
> > metastat did not work and the system asked me if I wanted to mount a
> > zpool which it found during startup. But there should not be any more
> > zpool now, all partitions are UFS.
>
> > The reboot did not work with c0t0d0s0 in the vfstab, so I rebooted once
> > again into failsafe. From this point, the whole system is corrupted. I
> > am unable to boot anymore. The OpenBoot prompt tells me this:
>
> > | Rebooting with command: boot
> > | Boot device: disk  File and args:
> > | ERROR: Last Trap: Fast Data Access MMU Miss
> > |
> > | {1} ok boot
> > | ERROR: /pci(a)1c,600000: Last Trap: Fast Data Access MMU Miss
> > |
> > | {1} ok
>
> > I am unsure how to proceed from this point. The second machine tells me
> > exactly the same error about "/kernel/drv/did.conf", so I do not think
> > that in both servers a hard drive failed simultaniously.
>
> > Regards
> >  Stephan
>
> Hi,
>
> not sure what you have done on those servers exactly - but the error
> you got (boot_archive and GRUB) is related to x86 architecture. but
> you have got v240 (SPARC).
>
> now, if I were in your shoes I would boot off OS CD or net into single-
> user mode and:
> * restore bootblk on the internal disks
> * mount your root FS from the internal disks and look into the
> following:
> /etc/system
> /etc/vfstab
>
> if you get the same error (MMU Miss) when booting off CD or net - open
> the call in Sun support.
>
> Cheers,
> Nikolai

ops, sorry - I was wrong - as I see it must be Solaris10 U6 - so, the
warnings to boot_archive can be met on this particular configuration
after patching - due to the boot_archive file is not updated after
patch has been applied.

anyway, before booting off CD or net - try to reset the server

from OBP
setenv auto-boot? false
reset-all

and then
try to boot off the local disk with -F failsafe option (I would try
from both disks), if it doesn't work - boot off cdrom or net (if you
have JS server)

if you still have MMU Miss - log the call to Sun

Cheers,
Nikolai
 | 
Pages: 1
Prev: Host power on is disabled due to:SCC is not valid or is not present
Next: Incoherent ssh daemon in Solaris 10!