 Questions and Answers
Amy Rich
Q I installed Solaris 10 on a Sun box that I had lying
around in order to investigate using ZFS. I chose to give ZFS entire disk
devices from an external disk pack instead of giving it individual disk
slices from each disk. After I was done playing, I wanted to re-install
Solaris 9 and slice up the disks, but it doesn't appear to work
anymore. Apparently ZFS deleted the disk label and I can no longer slice it
up for Solaris 9. Here's the command I ran to create the pool:
zpool create -f testpool mirror c2t0d0 c2t1d0
Now when I run format and try to partition the disk, it tells me:
Current partition table (unnamed):
Total disk sectors available: 143358320 + 16384 (reserved sectors)
Part Tag Flag First Sector Size Last Sector
0 usr wm 34 68.36GB 143358320
1 unassigned wm 0 0 0
2 unassigned wm 0 0 0
3 unassigned wm 0 0 0
4 unassigned wm 0 0 0
5 unassigned wm 0 0 0
6 unassigned wm 0 0 0
8 reserved wm 143358321 8.00MB 143374704
partition> 8
`8' is not expected.
Note that there is no slice 7 anymore. If I try to
partition slice 0 to start at sector 0, I get the following error:
partition> 0
Part Tag Flag First Sector Size Last Sector
0 usr wm 34 68.36GB 143358320
Enter partition id tag[usr]: root
Enter partition permission flags[wm]:
Enter new starting Sector[34]: 0
`0' is out of range.
Do I need to buy a new disk now that I've
modified this one, or is there some way to fix this?
A ZFS installed what's known as an EFI
(extensible firmware interface) label on your disk, replacing the previous
SMI label. See the following for more information on EFI labels and what
they're used for:
http://docs.sun.com/app/docs/doc/817-5093/6mkisoq1k?a=view
To begin, make sure you've removed your zpool
with zpool destroy testpool if you're still running under Solaris 10.
To rewrite the SMI label again, run the following command:
format -e
Choose the disk you want to modify and reset the
partition type to SMI:
format> partition
partition> label
[0] SMI Label
[1] EFI Label
Specify Label type[1]: 0
Warning: This disk has an EFI label. Changing to SMI label will erase
all current partitions.
Continue? y
Auto configuration via format.dat[no]?
Auto configuration via generic SCSI-2[no]? yes
When you look at the partition table now, you'll
see that it lists the SMI partitions 0 through 7 again. You can go back to
slicing your disk as you did previously.
Q I just upgraded a number of our departmental
Powerbooks to Mac OS X 10.4.7. Since the upgrade, users are complaining
that the Software Updater can't unpack software updates correctly. In
the log file of one such machine, I see the following errors:
Aug 28 06:13:58 mymac Software Update[27884]: JavaScript error
"Undefined value" while running "__choice_su_visible"
Aug 28 06:13:58 mymac Software Update[27884]: __choice_su_visible
returned error: Undefined value
If I download the update and apply it manually, it works just fine.
Google is unhelpful about how to correct this other
than "do an archive install instead of an upgrade", which is
completely unacceptable. Do you have any suggestions on how I might fix
this issue?
A Whenever there are issues with an OS X upgrade or
patch, I always suggest running the Disk Utility application to try to
"repair" permissions (this essentially does a chmod and chown on various files and directories) on the boot volume. If that doesn't
work, another step that commonly fixes issues is to reapply the Updater
that broke the behavior (in your case, presumably the 10.4.7 Updater).
I've successfully fixed the error you describe with those two steps.
Q I have a Solaris 8 machine that I'm trying to
bring to the ok prompt, but every time I try, it asks me for a password.
I've tried the root password, but that doesn't work. I'm
stumped as to what this password might be, so how do I get around it?
A You don't mention what kind of hardware or
exactly what the login message says, so there are a few possible scenarios.
You could have a firmware password set; you could have an LOM, ILOM, or
ALOM password set; or you could have an RSC password set. Each of these
requires a different method to circumvent.
Let's take the firmware password first, since
it's probably the easiest to deal with. Execute the following command
as root from the running system:
eeprom security-mode
If it does not say:
security-mode=none
then you have an eeprom password set, and you can
either turn down the security-mode to none or you can reset the password.
To turn off firmware security entirely, run:
eeprom security-mode=none
To instead change the password (security-mode must be
set to command or full for this to work), run:
eeprom security-password
It will prompt you for a firmware password twice.
If the firmware password is not your problem, then you
either have an ILOM/LOM/ALOM password or an RSC password, depending on the
type of hardware you're using. If you have a SunSolve account, look
at InfoDoc 81146 to determine which type of console management software you
have:
http://sunsolve.sun.com/search/document.do?assetkey=1-9-81146
If you have RSC, then you may need to install extra
software to obtain the binaries you need to reset the password. Follow the
directions at:
http://www.sun.com/servers/rsc_download_readme.html
to obtain and install the correct version of the RSC
software for your hardware platform and OS version. Once you have the
software installed correctly, run the following command to reset the admin
password:
rscadm userpassword admin
If it claims that the username does not exist
(possible if you just installed the software), then you can create a user,
set its password, and give it privileges:
rscadm useradd admin
rscadm userpassword admin
rscadm userperm admin cuar
The privileges set above are:
c - Console permission: needed to toggle between LOM and console \
(possible only when LOM and console share port A).
u - User accounts permission: needed to create/manage LOM user accounts.
a - Administrative permission: needed to re/set LOM
config variables.
r - Reset permission: needed to reset/power-cycle LOM.
If your hardware has ALOM, then the software you need
should already be installed as part of SUNWkvm. To reset the password, run
the following command:
/usr/platform/`uname -i`/sbin/scadm userpassword admin
If you did not have the root password, you could erase
all of the ALOM settings by following the instructions from:
http://www.sun.com/products-n-solutions/hardware/docs/html/ \
819-3250-11/trouble_appx.html#pgfId-1003331
If your hardware has LOM, the predecessor to ALOM,
then you'll need to install the LOMLite software available from the
Solaris 8 Supplemental CD. After inserting it, perform the following:
cd /cdrom/cdrom0/Lights_Out_Management_2.0/Product
pkgadd -d . SUNWlomm SUNWlomr SUNWlomu
You might then be able to bypass user security by
modifying /etc/lom.conf or /platform/sun4u/kernel/drv/lom.conf to disable serial_security:
serial_security=1;
If you're working with an older machine, you
might also have the option of setting a jumper on the motherboard to reset
the LOM entirely.
If you're using x64 hardware with ILOM and a
BIOS password, the default username is root and the default password is changeme. If it's been changed and you need to reset the password, it
requires physical access to the machine to open the case and modify the
physical jumper settings. Refer to your hardware guide to determine exactly
which jumper you need to change. For example, this document describes the
changes needed for the Sun Fire X4100 and X4200 class machines:
http://www.sun.com/products-n-solutions/hardware/docs/html/ \
819-1157-15/power-bios.html#pgfId-1001019
Q I have a Sun V240 running Solaris 9. I'm
currently using the default bge0 interface, but I want to replace it with a
quad gigE ce0 card. I added the card and plumbed the interface, and
everything looked good. I bought the interface up with a different IP and
things still looked okay. When I unplugged bge0, though, the machine
stopped responding. Here are the pertinent config files:
/etc/hosts:
127.0.0.1 localhost
10.1.1.1 host-bge0
10.1.1.2 host-ce0
/etc/hostname.bge0:
host-bge0
/etc/hostnmae.ce0:
host-ce0
When I run ifconfig -a, it shows everything up and
functioning as expected:
lo0: flags=1000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4> mtu 8232 index 3
inet 127.0.0.1 netmask ff000000
bge0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 4
inet 10.1.1.1 netmask ffffff00 broadcast 10.1.1.1.255
ether xx:xx:xx:xx:xx:xx
ce0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
inet 10.1.1.2 netmask ffffff00 broadcast 10.1.1.255
ether yy:yy:yy:yy:yy:yy
So why can't I get ce0 to function when I unplug bge0?
A Your netmask indicates that you're using a
class C-sized network, which means that the IP addresses for bge0 and ce0 are on the same subnet. This is an unsupported configuration unless
you're using IPMP. When you bring up ce0 after bge0, packets coming
out of the machine will leave the bge0 interface regardless of which IP
address was used by the other machine sending said packets. This is because bge0 is listed first in the routing table and is considered authoritative
for the 10.1.1.0/24 subnet. To verify this, you can look at the routing
table by doing netstat -nr.
The output should look something like the following
(with 10.1.1.254 assumed as the default gateway):
Routing Table: IPv4
Destination Gateway Flags Ref Use Interface
-------------------- -------------------- ----- ----- ------ ---------
10.1.1.0 10.1.1.1 U 1 4699 bge0
10.1.1.0 10.1.1.2 U 1 0 ce0
224.0.0.0 10.1.1.1 U 1 0 bge0
default 10.1.1.254 UG 1 31150
127.0.0.1 127.0.0.1 UH 4 705 lo0
If you run snoop -d bge0 and snoop -d ce0 at the same
time, you'll see packets coming into ce0 and responses to those
packets leaving via bge0.
If you do a flash cutover instead of trying to have
both devices up at the same time, ce0 should work just fine, assuming that
there aren't hardware problems. Remove /etc/hostname.bge0 and reboot
the machine. You may also run into an issue with the ARP cache on your
switch, so give it a few minutes before deciding it didn't work (or
clear the ARP cache if you have that capability).
Amy Rich has more than a decade of Unix systems
administration experience in various types of environments. Her current
roles include that of Senior Systems Administrator for the University
Systems Group at Tufts University, Unix systems administration consultant,
author, and charter member of LOPSA. She can be reached at: qna@oceanwave.com.
|