Help with Sol10 boot error

UNIX Solaris
1

I just ran the latest ( 9-27-11) solaris patch set on my sparc testbed ( ultra80 ). Brought down to single user mode and ran patches as I always have. When I rebooted it after completion it came up with the following error on boot.

"Fast Access Data MMU Miss".

I can boot to a cdrom in single user mode and mount all my partitions. I can see everything. I have checked all my variable and ran diags on eeprom and no problems show. I am thinking one of the patches changed the kernal or some config file pointing to the boot block. I don't see a super block error. I fsck'd the partitions and all was fine. I haev read over the patch readme and don't see anything that stands out specifically...

Any suggestions??? Thanks in advance for helpin the new guy on the forum!

2

Did you backup your box before you did the install?

3

lol.. unfortunately no... The one time I was in a hurry and didn't back it up. Luckily, my data is on different drives that can be mapped back.

---------- Post updated at 12:07 PM ---------- Previous update was at 12:02 PM ----------

My biggest concern is I have 7 more systems to patch and don't want to corrupt any of them. Would like to find the problem so I can correct it before it causes other problems.

TIA

4

Have you tried any of following: Doc ID 1004094.1

  • 1) Confirm that we are able to reach the OBP ( ok> prompt)
  • 2)Confirm probe-scsi-all is able to see the boot disk
  • 3) Confirm OBP parameter diag-switch is set to false
    [list]
  • Example output:
  • ok> printenv diag-switch
  • Diag-switch false
  • To correct the problem:
  • ok> setenv diag-switch false
    [/list]
  • 4) Using show-disks and devalias, output, verify that the boot-device alias matches a show-disks path.
  • 5) Following a reset-all, verify another boot attempt has failed while capturing script data.
5

Will look up the doc. Im guessing its a sun KB doc? I did find a few things on the oracle KB about errors on the latest patch. It goes back to patch 144500-19. I tried to back out the patch but it wont allow me. It does show in the logs where it created a new bootblock. I'm guessing that may be the problem. I just have not found how to reclaim the old bootblock. I did try recreating a new bootblock with intallboot from the cdrom. But that didn't work either...

answers below...

thanks!

6

Well I finally figured it out...

The latest cluster included a kernel patch. That patch 147440-02 was incompatible with any OBP prior to 4.12.1 and my old test bed was running 3.23. Booted to cdrom and mounted my drives and backed out the patch. Booted right up!