When I was interviewed for my first position as a network support engineer (front line) for Sun Services I was asked "What tools would you use to diagnose DNS issues?" To which of course I answered "dig, the Domain Information Groper". The guy looked at me blankly (I know his name) and said "No... what else? to LOOKUP"? To me there really wasn't anything else, or did he mean host(1)... I can honestly say I'd never heard of nslookup(1) but then he'd never heard of dig. But I knew I was right in my answer. I was astonished to find that Solaris didn't provide dig(1) out-of-the box then. I am happy to report that its available on all supported releases today.

What brought this memory to mind is the feeling that I got in my stomach then occurred again when I read the statement; "Once installed, T-patches do not show their "T" prefix in the output of showrev -p ... They never did". I'm certain circa 1998 'showrev -p' did just that. I even managed to find a single link which backed me up, http://ftp.lanet.lv/ftp/sun-info/sun-patches/106699.readme, also available from SunSolve. So what's the consensus here? Am I right or am I right?

A minor update to ut-where to add a '-V' option which causes ut-where to provide some additional verbosity in a zenity window. I can't see one wanting to use the option ongoing but its nice at first to help understand whats happening, for example when your screen does not automatically change because the utsettings are not set as you might expect them to be. Or because XRandR is not supported.

My Sun Ray TM session can now automatically switch desktop size when moving between different clients.

For example:

Having logged on originally at my desk, Sun Ray 2FS with a 1920x1200 monitor

I pull my card and move to a break area with a Sun Ray 150 with a much smaller screen

The whole desktop is displayed, no panning required

Sun Ray Server Software 4.1 includes a new standalone Xserver, Xnewt, based on Xorg 7.2 community source. Xnewt introduces the XRandR extension to Sun Ray which amongst other features enables the session to switch Video Resolution and resize the desktop dynamically. No more panning around when using on a smaller display, or unused space when moving to a larger display.

Using the example above, moving from the Sun Ray 2FS with a 1920x1200 monitor to a Sun Ray 150 xrandr reports:

$ /usr/X11/bin/xrandr -q
 SZ:    Pixels          Physical       Refresh
 0    640 x 480    ( 180mm x 135mm )
 1    800 x 600    ( 225mm x 169mm )
 2   1024 x 768    ( 289mm x 216mm )
 3   1152 x 900    ( 325mm x 254mm )
 4   1280 x 1024   ( 361mm x 289mm )
 5   1400 x 1050   ( 395mm x 296mm )
 6   1440 x 900    ( 406mm x 254mm )
 7   1600 x 1024   ( 451mm x 289mm )
 8   1600 x 1200   ( 451mm x 339mm )
 9   1680 x 1050   ( 474mm x 296mm )
 10  1920 x 1080   ( 542mm x 305mm )
*11  1920 x 1200   ( 542mm x 339mm )
Current rotation - normal
Current reflection - none
Rotations possible - normal 
Reflections possible - none

While by comparison the Sun Ray setting, as reported by utset is somewhat smaller:

$ /opt/SUNWut/bin/utset | grep '^Display Timing 1:'
Display Timing 1: 1024x768@60
$

Therefore I can dynamically change the desktop using xrandr command and shrink my desktop accordingly:

$ /usr/X11/bin/xrandr -s 1024x768

The one caveat at this time is that for xrandr to grow the desktop Xorg must have allocated sufficient memory for it. Obviously with the example above it would have done. But had I logged in on the Sun Ray 150 first there may have been a problem. Therefore you may need to set your preferred default resolution to the largest available value to allow for that growth. This is accomplished using utxconfig:

$ /opt/SUNWut/bin/utxconfig -r 1920x1200

So what of automatic switching?

Taking Chris' Sun Ray location idea a little further I re-wrote his script, ut-where, so that it automatically switches the display depending on the device setting. This modified script has a couple of other new features including

• Pops up a window asking for the location if it's not known using zenity(1) (times out after 5 mins if you don't answer).
• Sets your status as un-available if the location description includes the word "conference".
• May be used to start IM client and monitor clients using the -i option.
• As before sets your away status when you disconnect (remove your card), unless that is you have already set your status to away... Great when you want to set your away status to something descriptive before leaving.

I'm still toying with the idea of saving other preferences to the location file such as the preferred resolution and or pidgin status... Full options listed below.

$ ut-where -h
usage: ut-where [-vxX][[-i]|[-d]|[-s]|[-t]|[-l]|[-c]|[-a [description]]]

  -a	Add location, prompts for description if none supplied.
    	Colon characters in description are substituted by space character.
  -l	Display contents of location file.
  -i	Initialise; starts utaction monitors for connect/disconnect
    	Will also start IM client if none active for user.
    	Kills current "utaction ut-where" scripts
  -c	Update pidgin with current description:
    	This option is intended to be used by "utaction -c"
    	If description is unknown then a description dialogue is
    	requested using zenity.
    	If description contains "conference" then status is
    	set to "unavailable".
  -d	Update IM tool with away message (unless away status already set).
    	This option is intended to be used by "utaction -d"
  -t	Display current description.
  -v	Verbose mode
  -x	Debug
  -X	Debug functions

Without any options the appliance name and description are displayed.
$

Thus to start IM client and utaction monitors simply use ut-which -i.

One final note. I have found some devices have there default resolution set much lower than expected, i.e. 640x480. Therefore you may need to change those. The best way to do that is to use utsettings (press <shift>-<props> keys together) and change the resolution under the "Display" category.

Happy switching.

An upgrade to BIND 9.3.4-P1 is available for Solaris 10 Operating Environment by installing patch 119783-05 for SPARC or 119784-05 for x86 architectures.

BIND 9.3.4 provided a number of new features over BIND 8 which was supplied with the Solaris 8 and 9 Operating Environments. Additionally BIND 9.3.4 provides a number of compatibility features not available in BIND 9.2.4 (the FCS version of BIND in Solaris 10 OE). For details of all the changes refer to the migration notes.

Summary of Differences between BIND 9.3 from BIND 9.2

BIND 9.3 has a number of new features over 9.2 including:

• DNSSEC is now Delegation Signer (DS) based, RFC4033, RFC4034 and RFC4035. This collection of RFCs is otherwise known as DNSSEC-bis.

• DNSSEC Look-aside Validation (DLV) (experimental), RFC4431

• check-names is now implemented.

• rrset-order is more complete.

• IPv4/IPv6 transition support, "dual-stack-servers".

• IXFR deltas can now be generated when loading master files, "ixfr-from-differences".

• It is now possible to specify the size of a journal, "max-journal-size".

• It is now possible to define a named set of master servers to be used in masters clause, "masters".

• The advertised EDNS UDP size can now be set, "edns-udp-size".

• New Name Server SMF properties. With the introduction of BIND 9.3.4 new Name Server smf(1) properties have been introduced to provide an SMF compliant method for setting BIND 9.3.4 command line options. Changing the "start method" property of the BIND 9 service is no longer recommended.

For further information refer to the migration notes as above.

My friend and Colleague Isamu just asked me how I live-upgrade my home machines. Here's how:

Tunnel the NFS TCP port over ssh

Thanks to Chris Gerhard for this tip.

  $ ssh -fNCL 6049:enospc:2049  enoexec

where 'enospc' is the name of the nfs server that exports the install images and 'enoexec' is the name of a host that can access the nfs server which has AllowTcpForwarding set to "yes" in its /etc/ssh/sshd_config file.

Using nfs URLs mount the file system

   # mount nfs://localhost:6049/export/install /mnt

Remove Current lu packaging and add new packages, i.e.:

   # yes y | pkgrm SUNWluu SUNWlur SUNWlucfg
   ....
   # ls /mnt/i86pc/os/nv/
   69 68 70
   # luv=70
   # cd /mnt/i86pc/os/nv/$luv/Solaris_11/Product
   # yes y | pkgadd -d . SUNWlucfg SUNWlur SUNWluu
   ....
   # cd /

Upgrade

Firstly, get name of my current lu partitions using lustatus:

   # lustatus
   ....

Rename the one I'm replacing so that I know which version it is for now.

   # lurename -n snv_$luv -e ‹old BE›

Next use lumake to update older image with changes made in newer image, thanks to Brent Paulson for this tip. Obviously you can miss this step if you just created the image. This takes some time, it sends an email when its complete. Alternatively use lustatus to see when its done.

   # lumake -n snv_$luv -t now -m mail-address@domain

Finally, do the upgrade.

   # luupgrade -u -n ‹new BE› -s /mnt/i86pc/os/nv/$luv

Activate the new environment

Use luactivate and 'init 6' to reboot solairs, the reboot command will not suffice.

   # luactivate snv_$luv
   ....
   # umount /mnt
   # init 6 

Stace

Just a quick note. One other modification that I made to my xorg.conf file, which has possibly saved my marriage, was to have Xorg use the audio device for the default beep, nee bell. This is accomplished by simply adding the following to the InputDevice section:

   Option "BellDevice" "/dev/audio"

Thanks goes to "Ramblings" and to Alan Coopersmith for this handy tip.

Well as I commented previously the Window system, keyboard, USB mouse and integrated touch pad just worked on the Ferrari 4000 laptop without me changing a thing. The additional frkit enabled me to use the Multimedia keyboard and of particular note the four way scroll button.

Initially plugging in an external display caused my LCD screen to be switched off. Remembering that the BIOS had a "Power on Display" option I changed this from the default of "Auto" to "both" this then caused the LCD output to be duplicated on the external display. Great for running a projector or driving a second display at a more comfortable height and saving my back. But having two displays I would rather them display different information.

As before I started by looking to Google for xorg dual monitor

Very quickly it became apparent that I needed to know a few things about the hardware I was using. Thus I imagine that the details below are particular to the the Ferrari 4000, or at least computers with the same graphics card. But I hope to share with you some discoveries that may be of interest.

What graphics card?

The first issue then is "What graphics card does the laptop have?" I tried prtconf(1) initially before stumbling across scanpci(1M) which I happened to see Jon use:

# /usr/X11/bin/scanpci | grep X700
  ATI Technologies Inc Radeon Mobility X700 (PCIE)
# 

I've used grep above for brevity, the actual list I got was much longer. I then used Google to find out what each component listed was.

xorg.conf configuration

Having established that the laptop has an ATI card I add that to my Google search and, some hours later, I find what looks rather promising in a ubuntu forum entry by Ziox:

I thus made some minor changes to that to suit my configuration, the full section ended up as:

# from http://www.ubuntuforums.org/showthread.php?t=301961
# System Requirements:
#ONE Dual-Output ATI or Matrox Graphics card.
#A recent version of the free Xorg, without any binary graphics driver installed.
Section "Device"
Option "MonitorLayout" "LVDS,CRT"
Option "MergedFB" "true"
Option "OverlayOnCRTC2" "true"
# Physical location of secondary monitor in relationship to primary monitor
# Values of LeftOf, RightOf, Above, Below or Clone (as used for single screen).
Option "CRT2Position" "Above"
# Option "CRT2Position"	"Clone"
# MetaModes: list primary-secondary monitor resolutions:
# switchable with Ctrl-Alt-num-Plus and Ctrl-Alt-num-Minus.
# Home 19-inch LCD (GH19PS) resolution is 1280x1024:
# The office 24inch monitor can display upto 1920x1200, alas
# that does not seem possible here.  The best seems to be 1680x1050
# The first entry is the default. 
Option      "MetaModes"   "1680x1050-1280x1024 1680x1050-1680x1050 1400x1050-1280x1024 1280x1024-1024x768 1024x768-1024x768"
# As The LCD is always present, while the CRT is not, the LCD is the primary.
Option "MergedXineramaCRT2IsScreen0" "false"
Identifier  "Card0"
Driver      "ati"
VendorName  "ATI Technologies Inc"
BoardName   "Radeon Mobility X700 (PCIE)"
BusID       "PCI:1:0:0"
EndSection

Interestingly several of the options used above are not currently listed in the xorg.conf(4) manual page installed on my system. A Google search for the options didn't turn up much either. I presume the manual changes are still in the works. FYI, the manual pages for X11 commands are in /usr/X11/man so add that to your MANPATH or specify the -M option to man, for example "man -M /usr/X11/man xorg.conf".

The catch

When I shutdown and unplug the external monitor Xorg does not at the next boot compensate for the fact that it's not there any more. Meaning when I next switch the laptop on without the second monitor Xorg may still open windows out of sight. Looking at the Gnome Workspace Switcher applet one can see a representation of your new window, you just cant see it. the most recent window is likely to have focus (the window is selected and thus input and or X commands affect it) and you can thus use 'Alt-F7' to provoke "Move Window" and slide it in to view. Or indeed use 'Alt-Tab' to select the window and then 'Alt-F7' to drag it in to view. However as I'm not certain this will always work, and that I find it rather disconcerting I'd like to Xorg to reconfigure to a single display.

As Xorg does not currently seem to reconfigure itself in this scenario, I need to reconfigure xorg.conf myself for the single screen, logout and login. And all is well; Xorg restarts it's self when I log out and hence reads the new configuration.

Now as I currently use x11-ssh-askpass to read my Secure Shell password during login before any other windows are opened it makes sense, to me at least, to modify it to provide two extra buttons so that during login, when I would normally enter my SSH password, I can instead, if needs be, switch configurations and simply re-login again.

As a prof-of-concept I knocked up xchoice in dtksh and added it to my dtlogin(1X) config file $HOME/.dtprofile:

DTSOURCEPROFILE=true

# Concept.  Background it as X must start to show it.
$HOME/tools/sh/xchoice&

#
# Start SSH agent
#
if [ "$SSH_AUTH_SOCK" = "" -a -x /usr/bin/ssh-agent ]; then
#  pkill -u $LOGNAME ssh-agent
  eval `/usr/bin/ssh-agent`
  if [ -x $HOME/x11-ssh-askpass-1.2.4.1/x11-ssh-askpass ]; then
    SSH_ASKPASS=$HOME/x11-ssh-askpass-1.2.4.1/x11-ssh-askpass ssh-add < /dev/null
  else
    /usr/openwin/bin/xterm -geom +0-0 -e ssh-add
  fi
fi

It works well, but dtksh requires more than basic X support that does not seem to be available when x11-ssh-askpass is run, and thus the window does not appear until later when the full windowing system is up. Hence the requirement for something else. And this way I can provide two functions in one.

Having done all that I'm still interested to know if their is some other way to accomplish this without reconfiguration so please do comment if you know.

One final issue:

The external LCD not being of equal size pans left and right, which is fine, but it also pans the whole desktop up and down. I'd like to stop that if possible and thus prevent the Gnome panel (launch bar) from sliding of the bottom off the bottom of LCD.

Stace

So after the initial install I made the following initial tweaks to the installation.

• 1. Create user account.
• 2. Modify /etc/autohome
• 3. Change machine name (hostname)
• 4. Configure for UK keyboard
• 5. Add frkit
• 6. Add inetmenu (Internet Menu)
• 7. Update GNOME preferences and create (copy) user environment

I'm sure I'll be making some more in the near future but for now as this is a Developer release I have all the compilers and tools installed already which includes a recent enough version of XEmacs 21.4 to keep me happy for a while

Create user account.

Simply use the GUI as provided when root logged in or if you prefer use useradd(1M).

Modify /etc/autohome

As I'm used to using /home for the location of my home directory it makes sense for me to configure this. See automount(1M) for more information on this.

Change machine name (hostname)

solaris-devx is a nice name, but its time I re-awoke Scully.

Interesting point here was that previously to change the machine name there were several files that needed to be modified. Not so now, it seems all I have to do is change nodename(4) (that's /etc/nodename) and run uname(1M), or vice-versa:

uname -S scully
uname -n > /etc/nodename

NOTE: I'm not sure what affect, if any, that may have on your running services. Thus to be honest I simply set nodename and the name came into effect when I rebooted.

Note that if your DHCP server gives your machine a name then nodename is ignored (effectively running 'uname -S' for you on the name supplied by the DHCP server).

Configure for UK keyboard

A UK keyboard typically has a sterling pound sign (£) above the number while a US keyboard has what I call a hash or number sign, but what I have heard my American friends (and Audix) refer to as the "pound sign"! Initially I created an xmodmap(1) to map the correct keys but as Chris Gerhard points out I simply need to set an eeprom value:

# eeprom keyboard-layout=UK-English

However as I don't think that comes into effect until you reboot here is the xmodmap file and command to use it. Remember to backup the original first just in case!

$ xmodmap -pke > .xmodmap.backup
$ cat .xmodmap-ferrari4000
keycode  11 = 2 quotedbl
keycode  12 = 3 sterling
keycode  48 = apostrophe at
keycode  49 = grave notsign brokenbar
keycode  51 = numbersign asciitilde
keycode  94 = backslash bar brokenbar
$ xmodmap .xmodmap-ferrari4000

(the brokenbar and bar have been swapped around when compared to the physical keyboard, but believe me its better that way)

Update 25 Jan 2007: Use kbd(1) to set the keyboard (thanks to Alan Coopersmith for this tip).

Add frkit

The frkit , AKA the Ferrari kit (though it is not just for Ferrari's), adds some very useful extensions from the OpenSolaris laptop community:

* acpidrv

Adds support for the power button and battery statistics.

* gnome battery

A port of the GNOME battery meter

* powernow

Juggles the CPU speed based on demand.

* acerkb

Multimedia keyboard hack

* gnome emifreq

GNOME applet that displays current CPU state.

Further information on all of these features is available by executing 'frkit -s readme'.

One tip, when running frkit remember to set your http proxy if necessary (or indeed, unset it!):

# frkit
Could not download "http://opensolaris.org/os/community/laptop/downloads/frkit"
# http_proxy=webcache:8080 frkit
Probing acpidrv ... installed and up to date.

Probing gnome-battery ... installed and up to date.

Probing powernow ... installed and up to date.

Probing acerkb ... installed and up to date.

Probing gnome-emifreq ... installed and up to date.

No modules applicable to your system or none out of date.
#

Add inetmenu (Internet Menu)

The inetmenu is a cool little package which provides a command-line interface (CLI) or GUI to select different network configurations. Thus when for example the laptop is at home I simply select the DHCP-noNIS profile, and when in the office I select DHCP. The application makes the necessary configuration changes for the network access point the system is using.

The package installation output gave me the impression that I had to create some profiles in /etc/inetmenu but this is not the case as the defaults suffice for me. I did have to reboot though before it worked. I presume I had not started some service or something? After-all I've never rebooted Solaris in the past quiet so often

One tip, rather than editing user_attr(4) to add the Ginetmenu profile to your UID use the usermod(1M) command, but be careful to list all your non-default profiles though!

Update GNOME preferences and create (copy) user environment

Added the battery meter to gnome-panel and assigned some of the multimedia keys to gnome applications and/or functions.

As an old Solaris user I copied most of my shell scripts, user environment customizations (aliases, .profile and the like) across from my Sparc system. Perhaps I'll tell you more about those another day...

What's next?

The next thing at the top of my to-do list is to get the external display working at the same time as the built in one so I can avoid the hunched-back.

Stace

Tag: x86 Solaris

Once the Solaris OE installation had completed the install process installed Sun Studio 11 and relevant patches. No further intervention was required by me.

At the end of the installation Solaris ejected the DVD media and booted itself from the internal disk. I was then presented with a new GRUB menu allowing me too boot Windows or Solaris, the default. As Solaris began to boot after my first install there was a moment of doubt when the screen went blank and the following three lines were displayed:

SunOS Release 5.11 Version snv_55 64-bit
Copyright 1983-2006 SunMicrosystems, Inc.  All rights reserved.
Use is subject to license terms.

Then nothing for a while (around 1 to 2 minuets)... followed by a message indicating that the network interface had failed to allocate an address using DHCP! That's because I had not plugged in the RJ45 cable. I did not see this delay or get the warning message when I next booted with the cable attached.

Hostname: solaris-devx
Configuring devices.
Loading smf(5) service descriptions: 150/150
...

solaris-devx console login:

At this point the machine seemed to be waiting for me to login... But I know better and waited for the GUI login to appear... I waited... I'm sure I should wait... perhaps I need to choose X.Org or Xsun from some place? I wait some more. Some warning messages from sendmail and syslogd pollute my screen:

Jan 17 22:41:31 scully sendmail[1194]: [ID 702911 mail.crit] My unqualified host name (scully) unknown; sleeping for retry
Jan 17 22:42:31 scully svc.startd[7]: [ID 122153 daemon.warning] svc:/network/smtp:sendmail: Method or service exit timed out.  Killing contract 61.
Jan 17 22:42:31 scully svc.startd[7]: [ID 636263 daemon.warning] svc:/network/smtp:sendmail: Method "/lib/svc/method/smtp-sendmail start" failed due to signal KILL.
Jan 17 22:42:39 scully syslogd: line 45: WARNING: loghost could not be resolved

I know that I did not set a domain name and as such sendmail is not happy! I wait some more, around eight minutes in total. When my waiting is paid off when the login screen appears (I'll look into this and log a Change Request if necessary).

I login as root providing the password that I was asked for during the install and I choose gnome over CDE when asked for the windowing environment to use and I'm in.

The first thing to do is to add a local user. Thoughtfully the "User and Group " application has already been started for me.

added 19 Jan:

To prevent the messages above (but not necessarily cure the issue) I disabled sendmail (svcadm disable sendmail) and added 'loghost' to my loopback (127.0.0.1) entry in /etc/inet/hosts.

added 20 Jan:

I now find that the Ferrari can boot and provide me with the GUI login screen in less than 1 minute. So as suggested it might be that Solaris was simply busy updating bits in the background after the first few boots. After all I was installing different bits. As a brief test I re-enabled sendmail today but still find Solaris boots in less than 1 minute. Or perhaps its that I'm now using inetmenu?

Stace

Tag: x86 Solaris

The DVD boots to an initial GRand Unified Bootloader (GRUB) menu where I'm prompted to select one of the following boot options:

1. Solaris Developer Express 2. Solaris Express 3. Solaris Express Serial console ttya 4. Solaris Express Serial console ttyb (fir lx50, v60x and v65x)

The DVD spins some more and I'm presented with this menu

1. Solaris Interactive (default) 2. Apply driver updates 3. Single user shell

After selecting option 1. Java is setup, a windowing system is extracted and system identification begins....Shortly after which I'm presented with a warning that I must pay attention now as the next screen requires input from me within 30 seconds or I will face a non-GUI install. The system pauses at this point waiting for me to press Return...

A windowing systems presents a single window, or text widget, with the following:

Press return if you can read this

I hit return. Then select my language, English. Select Continent and Country, Europe / Britain (UK). I get other questions like these... When I realise that I'm using the touch pad on a graphical display, fantastic! Well, it is for me, As said back at the start I'm used to simply watching the system install its self over the network and or pressing function keys.

A little while later I get to a "Customize Fdisk Partitions" screen and I can see my FAT32LBA partition and the, newly named, "Solaris" partition.

This is followed, as I hoped, by the "Lay Out File Systems" window where I can now create Solaris mount points or Virtual disks within the fdisk (as created by parted partition). The install has offered me some defaults which I could simply accept.

However as I want to use live upgrade I created the following layout:

	Slice	File System	Size	Comment
	0	/	10000mb	Initial root
	1	/swap	2048mb	I hope for sys-suspend
	2	/root	10000mb	alternative root
	6	/opt	5000mb	Optional packages
	7	/export	30140mb	home

I only need half the space that I allocated to the root partitions and to /opt, but I'm being cautious. Incidentally I created a /opt as when I've live-upgraded my Sparc machine my /opt was getting left-out (or so I believe) But I'm not certain that I need it. No time to read up about that as I need to go and do other chores. Thus I set the install process off....

Stace

Tag: x86 Solaris

The CD download was so fast I have not looked at anything else in Richard Friedman's Resources for Running Solaris OS on a Laptop.

Having burnt the System Rescue CD and restarting windows (he he, little does it know); I stopped the Acer from booting and entered the BIOS setup by pressing F2. in the BIOS I used the cursor keys to skip along the top tabs to enter the boot screen. Where I then used the up and down cursor to select (highlight) the CD/DVD ROM and pressed F6 a number of times to make it the primary selection. Saved the selection with F10 and Acer started to boot from the CD image...

In next to no time, after selecting a UK keyboard from a list, I was presented with a root shell prompt above which was some useful information including this extract:

X.Org : You can use the graphical environment.  Type startx.
the graphical environment configuration is done automatically.
X.Org comes with Window-Maker and you can use several graphical tools:
- Partition manager:..gparted
- Web browsers:.......firefox-2.0 and dillo
- Text editors:.......gvim and leafpad

Note well that this version lists gparted and not QtParted. Without further-a-do I entered _startx_ and entered a big dark room! That is to say the screen went black. I gave it a short while and then held down the power key to reset everything (I tried control-c first to no effect).

At the next attempt I took the default keyboard (experience taught me to try defaults) but to no avail. I did see some warning messages fly up the screen but alas I could not make them out. I considered using a video camera to try and catch them but first I thought I would try parted (my assumption being that it may be a non graphical version).

(parted) print
Disk /dev/hda: 100GB
Sector size (logical/physical): 512B/512B
Partition table: msdos

Number	Start		End		Size		Type		File System	Flags
1			32.3kB	3150MB	3150MB	primary	fat32
2			3150MB	51.4GB	48.2GB	primary	fat32			boot, lba
3			51.4GB	100GB		48.6GB	extended					lba
5			51.4GB	100GB		48.6GB	logical	fat32

(parted) resize 2 3150MB 20GB
(parted) rm 3
(parted) print
Disk /dev/hda: 100GB
Sector size (logical/physical): 512B/512B
Partition table: msdos

Number	Start		End		Size		Type		File System	Flags
1			32.3kB	3150MB	3150MB	primary	fat32
2			3150MB	20.0GB	16.9GB	primary	fat32			boot, lba

So far so good... But then I got cocky! The 'help' showed a command named 'mklabel' and 'help mklabel' showed "mklabel LABEL-TYPE : create a new disk label (partition tabel)" and "sun" was listed as a valid type. So I entered 'mklabel sun'. The net result was that my working msdos label was over-written with a new blank partition table! Thankfully there is also a rescue command so I entered 'mklabel msdos' and then used 'rescue' to recover (at least that is what I thought) my previous partitions using the information above.

But how then should I proceed now? It occurs to me that I don't really know what Solaris is expecting or indeed wants from the partition table. I assume that Solaris will allow me to create logical partitions within a partition that I create for it... I really want to ring-a-friend but then that isn't in my rules of engagement so I'm just going to assume thats the case. But before I do anything else I'm going to boot windows to make sure its still happy:

(parted) quit
root % reboot
...
F2 (and reconfigure BIOS to boot from disk)
F10 (save config)
...

Nada, nothing, nil, except back-light blackness....

OK, so I'll have to start again... I rebooted with the System Rescue CD and setup the partition thus:

(parted) mkpartfs primary fat32 32.3kb 40GB
(parted) mkpart primary sun-ufs 40GB 100%
(parted) print
Disk /dev/hda: 100GB
Sector size (logical/physical): 512B/512B
Partition table: msdos

Number	Start		End		Size		Type		File System	Flags
1			32.3kB	40.0GB	40GB		primary	fat32			boot, lba
2			40.0GB	100GB		60GB		primary
(parted) quit
root % reboot

And I insert the Acer recovery DVD! Well, I thought I'd give that a go. The laptop booted a windows shell and the Acer recovery DVD asked for confirmation to restore the factory settings which I opted for. Alas the system failed to come back. I booted from another Windows XP DVD that I own into the recovery console and used diskpart, chkdsk, fixboot, fixmbr and exit... Ejected the DVD and Windows started its virgin voyage (it booted) Phew! Looking back I wonder if the rescue attempt before would have worked if I have tried fixmbr on that (as I believe that is what the problem was).

Next, I insert the Solaris DVD and reboot....

Stace

Tag: x86 Solaris

So without to much looking I very quickly discover many notes from other folks that have trodden this path before. Notably a number on Acer machines (no that was not a surprise to me).

In no particular order then:

Multi-Booting the Solaris 10 OS, Linux, and Microsoft Windows on a Laptop by Ifeyinwa Okoye

Interesting. I don't intend to use Linux at the moment, though I may try BrandZ later on. One thing I most certainly will do is provide an alternative Solaris partition so that I can continue to use live upgrade once the initial installation is complete. I also don't have Partition Magic and so intend to use an alternative product...

Richard Friedman's blog and specifically his Partitioned entry.

Same hardware. Same requirements and Same low-budget thoughts and link to System Rescue CD which includes QtParted; a Linux GUI disk partitioning tool.

Perhaps now is a good time to mention that currently Solaris does not include a non-Solaris disk partitioning tool (if you hadn't already guessed).

Richard has also put together a useful set of links on the BigAdmin site entitled Resources for Running Solaris OS on a Laptop. which I'm going to go and read now while the ISO image of System rescue CD v0.3.02 downloads...

Stace

Tag: x86 Solaris

I have worked with Solaris on Sparc for over twelve years, and while I did install Solaris 2.6_x86 on a home PC back in 1998 (or there abouts) I can honestly say I don't remember much about it. Applying of course that I do remember something and that something is booting from a floppy disk (there, I don't remember what that floppy was called) and having to configure the mouse and VDU via some configuration GUI thereafter installation. Then having installed it I never actually used it, as I had a Sparc 10 after-all. I must also confess that I have Solaris running on LX50's and v40's in the Lab. But for these machines I simply installed via jump-start and cared for nothing else except for nfs and ssh access which the jump-start took care of for me. Oh, and I sometimes use the x86 SunRay server....

So having got myself an Acer Ferrari 4000 (which doesn't have a floppy disk drive) I now intend to run the latest release of Solaris Nevada on it and share with you my experience as I do so. I also intend at first to limit myself to using only public information and to keep the pre-installed operating system intact. The caveat at this time is that I will use a yet unreleased version of Nevada which may soon be made available as a Solaris Express Developer release, as that is what I've been asked to install.

Thus I now start to gather information from blogs.sun.com and OpenSolaris.org...

Stace

Tag: x86 Solaris

I was forwarded by Jonathan to the History Matters One day in History link. I have included below my un-abridge entry, which alas was over the 3999 character limit. Enjoy:

A Day In History

My wife, Kirsty, and I awoke to Sara Kennedy on Radio 2 at 6:50am. We listened to Sara’s voice as she read out the headlines from today’s news papers; I don’t read a news paper my self as they have too much sensationalism and gossip to be worth the while in my opinion. We don’t lie listening in a slumber for to long, no Terry Wogan for us today. As Kirsty, a primary school teacher, is working as a class room assistant at Portesbery School, a special needs school, for the first time today.

Kirsty leaves the house at around 7:50. Emily, my eldest daughter leaves for Secondary school at 8:20 and I take my youngest daughter, Alice, to school by bicycle at 8:25am. It’s exactly a mile away and takes around 5 minutes to cycle there. I returned home to work rather than going in to the office today.

I work in Computing, I’m a senior software engineer for Sun Microsystems, specialising in naming protocols such as DNS (Domain Name system), LDAP (Lightweight Directory Access Protocol) and NIS (Network Information Protocol). In brief I diagnose issues and where necessary provide corrective actions, advice and or address the issue in the source code.

At home I work on a Sun Blade 150, a SPARC based processor designed by Sun. On which I run the current in-development version of the Solaris Operating Environment. I keep the system up-to-date with the latest development release using Live-Upgrade, which allows me to upgrade one disc partition while continuing to work on another one. This was the first thing I did today, before breakfast and making the pack-lunches I resumed Solaris, started a secure shell connection to the office, mounted the relevant NFS partition and started the upgrade process, luupgrade to build 50 of Nevada.

After returning from the school run (well, cycle) I started the online chat tool, Gaim, and email client, Thunderbird, and proceeded to wade through todays email. On average it takes two hours to work through the email. It’s a mixture of technical questions, bug updates, resource notifications, test results, third party developments, new projects, announcements… and much more. Email is the preferred communication tool and as such is revisited throughout the day. Almost everything is done electronically; I rarely receive post at work.

My plan for today was to investigate an issue with the DNS resolver; a software library which translates names to Internet Protocol addresses. The latest changes caused one or two other software applications to fail horribly. By which I mean an uncontrolled fault causing the operating system to stop the process and create a memory dump (core file), otherwise known as a crash. Thus I needed to analyse why Solaris seems to be mostly seeing this issue, look for other possible faulty applications (as not all are seeing this issue) and investigate if any changes could be made to the resolver to help guard against these faulty applications from crashing.

At lunch I cycled up to the local plumber merchants in search of a smaller u-bend. At the plumbers they have a knack for making one feel rather small if not inadequate; I’m not a plumber and am not sure if different size u-bends are available… I have been replacing our on-suit bathroom over some months and needed a different u-bend to fit the bidet; as the one I took off the other bidet would not fit in the space causing the bidet to be raised off the floor! Turned out that a standard bidet u-bend has a shorter neck for this very reason and I returned home triumphant.

Alas while I was away my Virtual Private Network tunnel to the office had gone down, reset, and vanished! Not a big issue, nothing is lost by this as I use a combination of software which is well suited for working in different locations. Namely, screen: for managing remote terminal sessions. And GNUEmacs: for file management, editing, text formatting, source code management, searching, just about everything else.

In the afternoon I worked some more on the libresolv issue. Documented my findings in the Change Request database , changed some libresolv source code which would prevent one of the crash scenarios and started the business of cross compiling for testing and verification.

Emily returned home around 2:40pm and began her homework by herself. I cycled back to school and collected Alice on my bike around 3:05pm. By 3:30 I was back in front of the terminal considering how I might address some of the issues that I had discovered today. I signed off just after 6pm, when I noticed that my live upgrade had stalled…. The secure shell (SSH) which I had NFS mounted the disc over would have been reset when the VPN tunnel was lost earlier. So rather than suspend the machine I invoked a new SSH session, NFS automatically recovered and subsequently live upgrade continued.

I joined my family for dinner and discussed my wife’s day at the special needs school. Seems she enjoyed it and is looking forward to working there again.

In the evening, around 7pm, I set off for a night time cycle with Jonathan and Keith, colleagues from work. We cycled up the Basingstoke Canal from Fleet towards Farnborough and headed in to the woods surrounding Tweseldown race course. We cycled around in the dark looking for single track routes, finding what we thought was one and then getting lost in trees and bushes. We thoroughly enjoyed ourselves and despite the number of brambles we encountered we all avoided a puncture. When I returned home I found that we were ridding in an area called Outridden Hill!

I returned home around 9:30pm and watched the final half hour of the excellent new BBC drama called “The Amazing Mrs Pritcahard”.

Before bed I returned to the computer, the live upgrade had completed. I selected the new OS and powered off the system; ‘init 5’.

Sun supported Solaris Operating Environments are not affected by the latest BIND vulnerabilities documented by the Internet Systems Consortium.

Solaris 8, 9 and 10 OE's are unaffected as they provide earlier versions of BIND which are not vulnerable to these latest vulnerabilities:

• SIG Query Processing (CVE-2006-4095):

Recursive servers:

Queries for SIG records will trigger a assertion failure if more than one SIG (covered) RRset is returned. Exposure can be minimised by restricting sources that can ask for recursion.

Authoritative servers:

If a nameserver is serving a RFC 2535 DNSSEC zone and is queried for the SIG records where the are multiple SIG(covered) RRsets (e.g. a zone apex) then named will trigger a assertion failure when it tries to construct the response.

• Excessive Recursive Queries INSIST failure (CVE-2006-4096):

It is possible to trigger an INSIST failure by sending enough recursive queries that the response to the query arrives after all the clients looking for the response have left the recursion queue.

Exposure can be minimised by restricting sources that can ask for recursion.

Note for BIND 9.2.x: Code handling this path for 9.2.x has been determined to be wrong, though ISC has not been able to detect an execution path that would trigger the erroneous code in 9.2.x.

Stace

Tag: BIND