From: Steve Howie (showie@uoguelph.ca)
Date: Sat Aug 21 2004 - 23:59:56 EDT
Howdy
We finished our Solaris 7 to 9 upgrade today, and while it worked for
99% of the tasks, there were a couple of very peculiar things I noticed
about couple of steps.
First, this was an INSTALL of Solaris 9 over the Solaris 7 system on our
Sun E450. This E450 has two external D1000 arrays, each on a separate
controller - controller C2 for the first and C3 for the second.
The Solaris 9 Installer, however, switched the controller numbers on me.
File systems which were mounted on C3 now had to be changed in vfstab to
be moiunted on C2. Any ideas why this would happen? It was running
happily on C2 and C3 for a number fo years, now they are switched.
Something to do with SCSI bus arbitration or something? Needless to say
this added a couple of hours to the overall task
Secondly, we had a couple of metadevices to recreate. After taking
Solaris 7 backups of the metadevice file systems along with paper
records of the config, I deleted the metadevices and database replicas.
After Solaris 9 was installed, the metadevices were manually recreated
with no problems at all. Except for 1 of the metadevices. This involves
7 LUNS from an HP XP512 Storage area network, all concatenated to give a
280Gb. file system. However, after acquiring and mapping the luns as OS
devices, concatenating them failed - metainit stated that the second
extent overlapped the metadevice. Which is, of course, nonsense.
This is an old error apparently, found back in SDS 4.2.1 days, and
appears to be going strong with Solaris 9 Volume manager. Apparently it
has something to do with the disk vendor ID's in non-SUN disks. The LUNS
on the HP show up as "HP-OPEN" disks: c4t0d14 thru c4t0d20 are the LUNS
from the XP 512 in the following 'format' output which I'm trying to
concatenate:
23. c3t13d0 <SUN18G cyl 7506 alt 2 hd 19 sec 248>
/pci@4,4000/scsi@2,1/sd@d,0
24. c4t0d14 <HP-OPEN-9*10-SUN-0118 cyl 50078 alt 2 hd 30 sec 96>
c4t0d14
/pci@4,2000/JNI,FCE@1/sd@0,e
25. c4t0d15 <HP-OPEN-9*10-SUN-0118 cyl 50078 alt 2 hd 30 sec 96>
c4t0d15
/pci@4,2000/JNI,FCE@1/sd@0,f
26. c4t0d16 <HP-OPEN-9*5-SUN-0118 cyl 50078 alt 2 hd 15 sec 96>
c4t0d16
/pci@4,2000/JNI,FCE@1/sd@0,10
27. c4t0d17 <HP-OPEN-9*5-SUN-0118 cyl 50078 alt 2 hd 15 sec 96>
c4t0d17
/pci@4,2000/JNI,FCE@1/sd@0,11
28. c4t0d18 <HP-OPEN-9*5-SUN-0118 cyl 50078 alt 2 hd 15 sec 96>
c4t0d18
/pci@4,2000/JNI,FCE@1/sd@0,12
29. c4t0d19 <HP-OPEN-9*6-SUN-0118 cyl 60094 alt 2 hd 15 sec 96>
c4t0d19
/pci@4,2000/JNI,FCE@1/sd@0,13
30. c4t0d20 <HP-OPEN-9*2-SUN-0118 cyl 20030 alt 2 hd 15 sec 96>
c4t0d20
/pci@4,2000/JNI,FCE@1/sd@0,14
Any idea how to fix? I did see a proposed solution which involved
padding the vendor id field on the disk with blanks - a bit ugly if u
ask me :)
Any help appreciated
Thanks!
Scotty
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:29:20 EDT