I have waiting been for more information in the hope that I could post a
definitive answer back to the list - alas, this is not to be.
Summary of original question (for the sake of completeness, the original
question is included below in its entirety) :
Ultra30, Solaris 2.6 server HW 3/98 with Symbios MetaStor Workgroup
attached. 'probe-scsi-all' causes the error "Fast Instruction Access MMU
Miss" or "Illegal Instruction" :
ok probe-scsi-all
/pci@1f,4000/scsi@5,1
/pci@1f,4000/scsi@5
Target 5
Unit 0 Disk SYMBIOS INF-01-00 0205
Unit 1 Device Type 20 SYMBIOS INF-01-00 0205
...
Unit 7 Device Type 20 SYMBIOS INF-01-00 0205
Fast Instruction Access MMU Miss
ok
Summary of replies :
Thank you to all who responded. One respondent wrote that he had
experienced the same errors on an E3000 (with 'boot net - install') but
had not found a explanation for the problem. Other respondents suggested
that the problem could be caused by :
- conflicting SCSI IDs [I have verified that all attached devices have
unique IDs]
- inability of U30 OpenBoot PROM (I have version 3.9) to "see" more than
certain number of disks / LUNs. The fact that probe-scsi-all always
bombs out after detecting Unit 7 certainly lends credibility to this
theory; however, I have not yet been able to find any verification of
this.
Responses from suppliers :
- Symbios suggested modifying the Workgroup startup options so that it
does not initiate the negotiation of SCSI protocols but instead waits
for host adapter to do so (apparently this can U30s to panic to boot
time). However, I don't think this is causing the symptoms I am seeing
and I have not attempted these changes.
- Sun suggested ensuring all latest patches were installed and also that
I consider upgrading the Flash PROM (Patch-ID# 105930-03). I am still
debating whether or not to do this (my problem does not appear in the
list of bugs fixed by this patch, so I wonder whether it is worth
messing around with it).
Current status :
My call with Sun is still open and I am awaiting further information /
instructions from them.
Best regards,
-- craig
Craig Faasen Email: craig.faasen@diasemi.de
System and Network Administrator Tel : +49 (0)7021 9414-40
Dialog Semiconductor GmbH Fax : +49 (0)7021 9414-10
http://www.diasemi.com whois: CF4-RIPE
Craig Faasen wrote:
>
> Dear Managers,
>
> I have a new Ultra30 with a (also new) Symbios MetaStor Workgroup RAID
> system attached to it :
> Ultra30, Solaris 2.6 server HW 3/98
> - 2 x internal 4.2 Gb disks
> - CDROM, floppy drive
> - X1032A SunSwift Ethernet/SCSI adaptor
> - X6541A UltraSCSI Differential adaptor (MetaStor is attached to this).
> MetaStor, 10 x 18Gb disks, RAID 5, 1 disk hot spare.
>
> When I do a 'probe-scsi-all', I get the following :
> ok probe-scsi-all
> /pci@1f,4000/scsi@5,1
> /pci@1f,4000/scsi@5
> Target 5
> Unit 0 Disk SYMBIOS INF-01-00 0205
> Unit 1 Device Type 20 SYMBIOS INF-01-00 0205
> Unit 2 Device Type 20 SYMBIOS INF-01-00 0205
> Unit 3 Device Type 20 SYMBIOS INF-01-00 0205
> Unit 4 Device Type 20 SYMBIOS INF-01-00 0205
> Unit 5 Device Type 20 SYMBIOS INF-01-00 0205
> Unit 6 Device Type 20 SYMBIOS INF-01-00 0205
> Unit 7 Device Type 20 SYMBIOS INF-01-00 0205
>
> Fast Instruction Access MMU Miss
> ok
>
> Sometimes the error is "Illegal Instruction" but it always appears after
> "Unit 7" above.
>
> With the MetaStor powered down, 'probe-scsi-all' finds all attached
> devices
> without any problem :
> ok probe-scsi-all
> /pci@1f,4000/scsi@5,1
> /pci@1f,4000/scsi@5
> /pci@1f,4000/scsi@3
> Target 0
> Unit 0 Disk SEAGATE ST34371W SUN4.2G7462
> Target 1
> Unit 0 Disk SEAGATE ST34371W SUN4.2G7462
> Target 6
> Unit 0 Removable Read Only Device TOSHIBA XM5701TASUN12XCD2395
> /pci@1f,4000/pci@2/SUNW,isptwo@4
> ok
>
> With the Metastor powered up, I have tried :
> - 'test-all' - no problem
> - 'sunvts' on all devices - no problem
> - 'newfs' and mount the RAID - no problem.
> - 'health check' using the RAID configuration software - 'No
> exceptions'.
>
> Despite all the above 'no problems', I'm naturally concerned that
> whatever is causing the 'probe-scsi-all' problem may rear its ugly head
> when the server is in production. I've had a search through the archives
> and various Sun sites, but could not find any conclusive explanations.
> Has anybody experienced this or have any ideas on how I can isolate what
> is causing the error.
>
> Best regards,
>
> -- craig
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:43 CDT