After having Ubuntu upgrade leave my netbook useless (I had to re-install), and - since this happened on the road when I really could have used my machine - I was determined (after doing a clean-install of Ubuntu 9.10) to have alternate boot options on my netbook.
At least I'd have a way to keep working.
I also wondered if any of the existing installs would properly manage the touchpad (that is, allow disabling while typing).
I decided to install a small image of the original windows, an Ubuntu and a Fedora.
I set about to create various test live-images on USB so I could test what worked well, and then work out the backup and installation sequence I would need.
I started from a Fedora boot on my desktop. Fedora images are created by Live USB Creator. I could create other test images from Fedora with UNetBootin.
I had been using UNetBootin for a while - it seems pretty nice, automatically downloading a distro (or taking a path), and being pretty smart about selecting the USB it would configure.
I worked back and forth between Live USB Creator and UNetBootin. At one point, I was testing a release candidate F12, and quickly used Live USB Creator (after using UNetBootin for the last several creations, including some USB ones).
Ach! Live USB Creator picked a disk with my home directories, and DELETED user areas before I realized what it was doing. VERY NASTY.
Beware!
Ubuntu 9.04 upgrade quashes audio
The other night, as I usually do, I accepted a slew of Ubuntu updates. My audio was gone afterwards.
After some searching, this thread proved helpful: http://ubuntuforums.org/showthread.php?t=205449 but I did need to figure out what an appropriate ASLA driver would be.
The soundcard list on the ASLA site showed two options; I went for the HDA one, since
I followed the instructions on the ubuntuforums link to build from sources, compiling the snd-hda-intel driver with the help of module-assistant (building couldn't have been easier).
The volume control panel has trouble opening (I'm not sure if this is because of the debug build). Basic sound works, but my Logitech headset is no longer seen, and so my skype is out of commission (does not make me happy - my phone of preference suddenly was made inoperable!).
UPDATE:
It appears that the generic 10de NVIDIA HDA (OSS mixer) sound device, if selected from the volume control hangs the volume control - and you can (?) never get back to it. The applet is gnome-volume-control: and when it hangs, you can't even select another window, do anything at all. CTRL-ALT-F1 to get to a normal terminal login allows you to find and kill that process. Then you can get back to the X-Win screens (CTRL-ALT-F7).
I tried installing from the Ubuntu distribution (as per the Ubuntu Forums); I tried building from the alsa sources from Ubuntu; I tried building from the ALSA latest sources (all drivers) from the ALSA-Project. I tried both before, and after (repeatedly) removing pulseaudio. I tested, looked thru the lsmod list on my netbook (Ubuntu 9.10) install.
As one last ditch, I'm going to try to remove what the Ubuntu Forums suggest, get it back to distro with apt-get, and then build the entire distro ASLA sources, and see what that gets me.
It looks like pulseaudio doesn't know what to do. I have rhythmbox playing music through speakers, while I try a test call with skype. I can initially succeed in having the outputs separate, but I get no input through the skype app from the headset microphone. If I try, the music shifts into the headset (out of the speakers) and, on a song transition, back again.
Finally I'll try what I posted here in the first place about removing pulseaudio:
UPDATE:
Well, I have restored Skype (more fundamentally, usb Logitech headset recognition on this installation) - this seemed to be accomplished by building / installing _all_ the sound drivers.
For future fallbacks, I can (still) use skype on my netbook (Ubuntu 9.10), or my Fedora-11/64 install.
This was way more work than I would ever want a normal system update to cause me. Perhaps its time to think of a simple mercurial versioning of update actions, perhaps limited to the last 10 revisions (or so) - a rollback system like, for example, Windows has.
Anybody up for designing this for the Linux community?
After some searching, this thread proved helpful: http://ubuntuforums.org/showthread.php?t=205449 but I did need to figure out what an appropriate ASLA driver would be.
The soundcard list on the ASLA site showed two options; I went for the HDA one, since
lspci -vshowed HDA from the NVIDIA driver on my ASUS board, that is to say I needed the snd-hda-intel driver. Not intuitive, but it worked... sort of...
I followed the instructions on the ubuntuforums link to build from sources, compiling the snd-hda-intel driver with the help of module-assistant (building couldn't have been easier).
The volume control panel has trouble opening (I'm not sure if this is because of the debug build). Basic sound works, but my Logitech headset is no longer seen, and so my skype is out of commission (does not make me happy - my phone of preference suddenly was made inoperable!).
UPDATE:
It appears that the generic 10de NVIDIA HDA (OSS mixer) sound device, if selected from the volume control hangs the volume control - and you can (?) never get back to it. The applet is gnome-volume-control: and when it hangs, you can't even select another window, do anything at all. CTRL-ALT-F1 to get to a normal terminal login allows you to find and kill that process. Then you can get back to the X-Win screens (CTRL-ALT-F7).
I tried installing from the Ubuntu distribution (as per the Ubuntu Forums); I tried building from the alsa sources from Ubuntu; I tried building from the ALSA latest sources (all drivers) from the ALSA-Project. I tried both before, and after (repeatedly) removing pulseaudio. I tested, looked thru the lsmod list on my netbook (Ubuntu 9.10) install.
As one last ditch, I'm going to try to remove what the Ubuntu Forums suggest, get it back to distro with apt-get, and then build the entire distro ASLA sources, and see what that gets me.
cd /usr/src; sudo tar xjvf alsa-driver.tar.bz2; cd modules/alsa-driverI thought about removing a little more than the Ubuntu Forums say - e.g. libasound2 and libasound2-plugins - but apt-get pulled far too many libraries into the mix with that - close to 300. For now, I'll just try the standard thing the forum suggests:
sudo ./configure --with-kernel=/usr/src/linux-headers-$(uname -r) --with-cards=all --with-isapnp=yes --with-sequencer=yes --with-oss=yes
sudo make
Next, I'll install the ASLA driver as I built it over this:sudo apt-get --purge remove linux-sound-base alsa-base alsa-utils
sudo apt-get install linux-sound-base alsa-base alsa-utils
sudo apt-get install ubuntu-desktop
and reboot...sudo make install
It looks like pulseaudio doesn't know what to do. I have rhythmbox playing music through speakers, while I try a test call with skype. I can initially succeed in having the outputs separate, but I get no input through the skype app from the headset microphone. If I try, the music shifts into the headset (out of the speakers) and, on a song transition, back again.
Finally I'll try what I posted here in the first place about removing pulseaudio:
killall pulseaudioIt looks like this works. There is a bit of a caveat: the pulseaudio server seems to be running; I had to su (sudo was not sufficient) to stop it (pulseaudio --kill). It looks like I'd created a .asoundrc file which started up pulseaudio; if you have one, you will want to comment out any pulseaudio lines in it, or (as in my case) remove the file (pulse was the only stuff in it). Everywhere in /etc it is set to NOT start on startup, but it is there, and can run, for example, when fired by the menu item Applications->Sound & Video->PulseAudio Volume Control. It may be useful to remove these applications
sudo aptitude remove pulseaudio
sudo aptitude install esound
sudo rm /etc/X11/Xsession.d/70pulseaudio
UPDATE:
Well, I have restored Skype (more fundamentally, usb Logitech headset recognition on this installation) - this seemed to be accomplished by building / installing _all_ the sound drivers.
For future fallbacks, I can (still) use skype on my netbook (Ubuntu 9.10), or my Fedora-11/64 install.
This was way more work than I would ever want a normal system update to cause me. Perhaps its time to think of a simple mercurial versioning of update actions, perhaps limited to the last 10 revisions (or so) - a rollback system like, for example, Windows has.
Anybody up for designing this for the Linux community?
HP mini 311
I wanted long battery life (something I could write with on trips, the train), and something to run Linux on. Most of the netbooks out this summer (2009) had screen resolutions which didn't suite them for making presentations at meetings, conferences, etc. Even worse, they seemed to be most usable with so called "Netbook Remixes" of Linux (e.g. Ubuntu, Fedora) which tried to make presentations more readable on the small, low resolution screens.
The 311 offered an NVidia chipset, an HDMI output port, and better resolution, while still holding reasonable battery life (and - at least for my preferences - the HP had a usable keyboard).
To save money, I got a basic, in-stock version from Amazon (no shipping costs, etc. made that the best cost option).
I had a Hitachi Travelstar 7200 320G drive, which happens to be conservative for standby power (better than the 160G drive that came with the 311). I added 2G RAM, which brought the total RAM up to 3G.
Access to the drive and RAM areas couldn't be nicer - see http://www.mobiletechreview.com/notebooks/HP-Mini-311.htm for some nice pictures.
I started out with a clone of my Ubuntu 9.04 system (a dd copy) - it seemed to work and update fine, but the touchpad would not be recognized by the mouse driver. What I mean is the touchpad worked fine, but there was no "touch" controls in Ubuntu, so that tap-to-click was always active, and made any typing a significant challenge (as the cursor would jump about quite readily... really, it's a great touchpad). In Windows, I was able to disable the tap-to-click feature, and then could both navigate and type usefully.
I tried all the "regular" suggestion on the web to get the touchpad recognized, but to no avail. On one trip, I let Ubuntu Jaunty (9.04) "upgrade" overnite, hoping this might help. It made my system totally unusable (wouldn't even boot - no keyboard; a real disappointment on a meeting trip!). When I got home, I did a "fresh" install of Ubuntu 9.10, and that is what I run now - it worked fine, but did not help with the touchpad.
Finally, I found a workaround here: http://ubuntuforums.org/showthread.php?t=1316361
I added these aliases:
The above mentioned thread seems to now have some working / hacked version of psmouse (with added hardware identifier), so no doubt there will be better solutions soon. For now, the aliases give me the control I want, without any worries about upgrade conflicts or problems.
The 311 offered an NVidia chipset, an HDMI output port, and better resolution, while still holding reasonable battery life (and - at least for my preferences - the HP had a usable keyboard).
To save money, I got a basic, in-stock version from Amazon (no shipping costs, etc. made that the best cost option).
I had a Hitachi Travelstar 7200 320G drive, which happens to be conservative for standby power (better than the 160G drive that came with the 311). I added 2G RAM, which brought the total RAM up to 3G.
Access to the drive and RAM areas couldn't be nicer - see http://www.mobiletechreview.com/notebooks/HP-Mini-311.htm for some nice pictures.
I started out with a clone of my Ubuntu 9.04 system (a dd copy) - it seemed to work and update fine, but the touchpad would not be recognized by the mouse driver. What I mean is the touchpad worked fine, but there was no "touch" controls in Ubuntu, so that tap-to-click was always active, and made any typing a significant challenge (as the cursor would jump about quite readily... really, it's a great touchpad). In Windows, I was able to disable the tap-to-click feature, and then could both navigate and type usefully.
I tried all the "regular" suggestion on the web to get the touchpad recognized, but to no avail. On one trip, I let Ubuntu Jaunty (9.04) "upgrade" overnite, hoping this might help. It made my system totally unusable (wouldn't even boot - no keyboard; a real disappointment on a meeting trip!). When I got home, I did a "fresh" install of Ubuntu 9.10, and that is what I run now - it worked fine, but did not help with the touchpad.
Finally, I found a workaround here: http://ubuntuforums.org/showthread.php?t=1316361
I added these aliases:
alias touchoff='sudo modprobe -r psmouse'This successfully disables the touch-pad altogether, but leaves USB mice working fine. I added a little IOGear Bluetooth 2.1 USB Micro Adapter (model GBU421; I got mine at CDW). My favorite mouse worked right away.
alias touchon='sudo modprobe psmouse'
The above mentioned thread seems to now have some working / hacked version of psmouse (with added hardware identifier), so no doubt there will be better solutions soon. For now, the aliases give me the control I want, without any worries about upgrade conflicts or problems.
Skype on Ubuntu Jaunty
I wanted to get skype up on my box; more and more people from open source forums around the world are on skype, and sometimes it's just easier to talk than IM.
I picked a USB headset from Logitech (ClearChat Comfort USB), and downloaded the newest Skype Beta (2.1 - Sept. 2009) For Ubuntu 8.10+, 32 bit (I'm running a 32 bit Ubunty, a 64 bit Fedora).
I could not seem to get skype to send sound out to the USB headset, so after some stumbling around forums and various suggestions, here's what worked for me.
As per the Troubleshooting Skype section on ubuntu.com/community/skype, I followed the suggestion (for Intrepid/8.10 there) to remove pulseaudio and install esound.
This seemed to work at first - I got menu choices in the skype options->sound devices page, and was able to get sound out of the headset, but getting recording to work (and trying all the device options) was not working.
I'd checked the volume control to make sure the record levels were set and not muted for the Logitech USB headset. I set System->Preferences->Sound to Sound Capture from the Logitech USB Headset (ALSA) - I left the rest of the device settings at Autodetect. I decided to try the gnome-sound-recorder (applications->sound & video->sound recorder), and that pointed me to which of the Logitech listed devices worked. I tried this in skype (no luck). I then tried recording with Audacity, and that gave me the needed hint: First, I had just one device (the same one) set for both playback and record in Audacity, yet it failed to record. Changing the settings from stereo to mono fixed that. The hint being - same device, same channel in skype - and that did the trick.
I made a test call, and then a test call to a "real" phone - report of sound was "better than any other phone in my home". Ok. Skype, here I come.
I picked a USB headset from Logitech (ClearChat Comfort USB), and downloaded the newest Skype Beta (2.1 - Sept. 2009) For Ubuntu 8.10+, 32 bit (I'm running a 32 bit Ubunty, a 64 bit Fedora).
I could not seem to get skype to send sound out to the USB headset, so after some stumbling around forums and various suggestions, here's what worked for me.
As per the Troubleshooting Skype section on ubuntu.com/community/skype, I followed the suggestion (for Intrepid/8.10 there) to remove pulseaudio and install esound.
killall pulseaudio
sudo aptitude remove pulseaudio
sudo aptitude install esound
sudo rm /etc/X11/Xsession.d/70pulseaudioThis seemed to work at first - I got menu choices in the skype options->sound devices page, and was able to get sound out of the headset, but getting recording to work (and trying all the device options) was not working.
I'd checked the volume control to make sure the record levels were set and not muted for the Logitech USB headset. I set System->Preferences->Sound to Sound Capture from the Logitech USB Headset (ALSA) - I left the rest of the device settings at Autodetect. I decided to try the gnome-sound-recorder (applications->sound & video->sound recorder), and that pointed me to which of the Logitech listed devices worked. I tried this in skype (no luck). I then tried recording with Audacity, and that gave me the needed hint: First, I had just one device (the same one) set for both playback and record in Audacity, yet it failed to record. Changing the settings from stereo to mono fixed that. The hint being - same device, same channel in skype - and that did the trick.
I made a test call, and then a test call to a "real" phone - report of sound was "better than any other phone in my home". Ok. Skype, here I come.
Firefox and Advertising sites...
I have noticed my browser becoming sluggish, bordering on inactive - this is Firefox 3.5.2 (mostly I notice it on Ubuntu's Firefox, but this is confirmed, definitely true for Fedora's Firefox too - worse: it seems to be doing lots of disk activity, slowing down the entire pc; killing intellicast.com site restores sanity...).
It's subtle - at first I thought it might be a memory leak, or a loop in Firefox, but I noticed that the weather page I usually keep open (intellicast.com) - if I kill the page, the responsiveness comes back, and the system monitor shows CPU cycles (pegged when this happens) fall immediately. It doesn't happen when you open intellicast; sometimes it happens in a few minutes (~15); sometimes after an afternoon. Because of this, I suspect it has something to do with javascript from an advertiser site (rather than something in the core intellicast site).
It's annoying. It's beyond annoying...
It's subtle - at first I thought it might be a memory leak, or a loop in Firefox, but I noticed that the weather page I usually keep open (intellicast.com) - if I kill the page, the responsiveness comes back, and the system monitor shows CPU cycles (pegged when this happens) fall immediately. It doesn't happen when you open intellicast; sometimes it happens in a few minutes (~15); sometimes after an afternoon. Because of this, I suspect it has something to do with javascript from an advertiser site (rather than something in the core intellicast site).
It's annoying. It's beyond annoying...
Ubuntu Kernel Update & NVIDIA Driver
Earlier, I reported that Fedora failed miserably on booting w/ a new kernel with the NVIDIA graphics driver installed (which I want, for two-headed desktop).
I hadn't seen a problem with Ubuntu Kernel update, but that appears to only have been true for one update - but not the latest. The good news is Ubuntu detects and reports that NVIDIA complains, and nicely defaults to the default driver (and single-headed mode).
Since, effectively, Ubuntu (and it appears Debian in general) does not have a non-graphic multi-user mode akin to init 3, here's what I did:
UPDATE:
As of kernel 2.6.28.15, it doesn't appear that gdm stop works (no terminal). This, however, seems to work for me:
UPDATE:
If running under telinit 1, NVIDIA driver will complain about potentially not being able to find files when searching in "init 1" - continue anyway (telinit 1 is not quite init 1 on other systems). When coming back up, you can telinit 3. I've found external USB hubs (e.g. printers, drives) are not re-initialized, so I prefer to reboot (but you can quickly telinit 3 to check video settings first, reboot later).
--------
If you came up graphical in Ubuntu before you discovered a problem w/ the kernel interface for the NVIDIA driver, then Ubuntu changed your config. When you come up (gdm start, or reboot if you like) then run the NVIDIA settings program to configure your monitors as you like as root:
You need to do this as root to save the configuration for future boots.
I hadn't seen a problem with Ubuntu Kernel update, but that appears to only have been true for one update - but not the latest. The good news is Ubuntu detects and reports that NVIDIA complains, and nicely defaults to the default driver (and single-headed mode).
Since, effectively, Ubuntu (and it appears Debian in general) does not have a non-graphic multi-user mode akin to init 3, here's what I did:
- Be sure to have _some_ nvidia driver available;
- exit Gnome window manager:
- sudo /etc/init.d/gdm stop
UPDATE:
As of kernel 2.6.28.15, it doesn't appear that gdm stop works (no terminal). This, however, seems to work for me:
- sudo telinit 1
- (login as root)
Alternately, if booting (Ubuntu, grub2), and you want to come up with the boot menu, then right-shift should get you there: pick recovery mode, and boot into shell to accomplish the following.
--------- run the NVIDIA run script with the update option; this will get a newer driver from the NVIDIA ftp site if one is available:
- sudo sh NVIDIA*.run --update
UPDATE:
If running under telinit 1, NVIDIA driver will complain about potentially not being able to find files when searching in "init 1" - continue anyway (telinit 1 is not quite init 1 on other systems). When coming back up, you can telinit 3. I've found external USB hubs (e.g. printers, drives) are not re-initialized, so I prefer to reboot (but you can quickly telinit 3 to check video settings first, reboot later).
--------
If you came up graphical in Ubuntu before you discovered a problem w/ the kernel interface for the NVIDIA driver, then Ubuntu changed your config. When you come up (gdm start, or reboot if you like) then run the NVIDIA settings program to configure your monitors as you like as root:
- sudo nvidia-settings
You need to do this as root to save the configuration for future boots.
VirtualBox for Smart Phones
There are more and more smart phones with linux kernels in their base. These are quickly going to become search / info / utility appliances and, I predict, replace a significant part of everyday personal computer use (you will still want a computer to do work, but for messaging, searching, simple connectivity that is / will be overkill, even in the face of netbooks).
It follows that the development process, then, of web applications (and other utility applications) will need to become less (not more) specialized. The good news - this will slowly come. The beginnings are starting to show up.
Android Phone "Live CD", usable in VirtualBox;
Palm PRE SDK (also using VirtualBox);
The Blackberry SDK - J2ME development (Eclipse); currently their tool say they require windows for their MDS RAD tool;
iPhone SDK - Requires Mac OS/X, and last I read approval from Apple before you can distribute your apps;
As you can see, this leaves you with 4 different development approaches, the last two providing their own emulators. The first three have plugins for Eclipse (so that is at least some common development feel). The first two can use Linux for a development environment (in fact, Android's "standard" development environment has been Ubuntu). The Blackberry development environment seems to be Windows only, but you _could_ run that under virtualbox on Linux, possibly with the Windows-7 Release Candidate.
So at the root, the common environment seems to be a small overlap of Palm-Pre & Android, although I don't think the development approaches are similar. In short, providing seemless services for the new "mobile pocket PC" (e.g., smartphones) is a steep hill. Clearly, this will have to change at some point. Right now, there seem to be 4 competitors for this space. A similar shootout is taking place for RIA client end, with Flex/Flash vs. JavaFX (?) vs. Json (is SVG anywhere in this mix?). That's a lot of choices for a developer of end-to-end solutions for this space.
One interesting item: Android has a native scripting environment, which (among other things) gives you access to phone facilities through a natively executed Python interpreter.
More on this later.
It follows that the development process, then, of web applications (and other utility applications) will need to become less (not more) specialized. The good news - this will slowly come. The beginnings are starting to show up.
Android Phone "Live CD", usable in VirtualBox;
Palm PRE SDK (also using VirtualBox);
The Blackberry SDK - J2ME development (Eclipse); currently their tool say they require windows for their MDS RAD tool;
iPhone SDK - Requires Mac OS/X, and last I read approval from Apple before you can distribute your apps;
As you can see, this leaves you with 4 different development approaches, the last two providing their own emulators. The first three have plugins for Eclipse (so that is at least some common development feel). The first two can use Linux for a development environment (in fact, Android's "standard" development environment has been Ubuntu). The Blackberry development environment seems to be Windows only, but you _could_ run that under virtualbox on Linux, possibly with the Windows-7 Release Candidate.
So at the root, the common environment seems to be a small overlap of Palm-Pre & Android, although I don't think the development approaches are similar. In short, providing seemless services for the new "mobile pocket PC" (e.g., smartphones) is a steep hill. Clearly, this will have to change at some point. Right now, there seem to be 4 competitors for this space. A similar shootout is taking place for RIA client end, with Flex/Flash vs. JavaFX (?) vs. Json (is SVG anywhere in this mix?). That's a lot of choices for a developer of end-to-end solutions for this space.
One interesting item: Android has a native scripting environment, which (among other things) gives you access to phone facilities through a natively executed Python interpreter.
Another Fedora Kernel Update
I followed suggestions on the Fedora forums - now when rebooting after a kernel update, the fastest way to re-run the NVIDIA driver install (and thus a recompile of kernel interfaces):
- at grub boot menu, "a" to alter kernel options;
- add '3' (without quotes) to the end of the line to come up init 3
- login as root, and run the NVIDIA installer (with the --upate option, it will check if a newer NVIDIA driver is available from the nvidia ftp site, and get it)
- reboot (actually, could probably just init 5...)
Fedora Library Conflicts
Today, I attemped a recommended system update on Fedora-11.
I had a conflict between a i586 and an x86_64 qt library. Not knowing where the 586 library came from, I yum remove'd it - it removed the following dependencies:
I guess with 64-bit Fedora, you have to find a 64-bit skype (?).
Not sure.
I had a conflict between a i586 and an x86_64 qt library. Not knowing where the 586 library came from, I yum remove'd it - it removed the following dependencies:
Running TransactionThe system update then worked.
Erasing : 1:qt-4.5.1-13.fc11.i586 1/5
Erasing : 1:qt-x11-4.5.1-13.fc11.i586 2/5
Erasing : phonon-backend-xine-4.3.1-6.fc11.i586 3/5
Erasing : phonon-4.3.1-6.fc11.i586 4/5
Erasing : skype-2.0.0.72-fc5.i586 5/5
Removed:
qt.i586 1:4.5.1-13.fc11
Dependency Removed:
phonon.i586 0:4.3.1-6.fc11 phonon-backend-xine.i586 0:4.3.1-6.fc11
qt-x11.i586 1:4.5.1-13.fc11 skype.i586 0:2.0.0.72-fc5
I guess with 64-bit Fedora, you have to find a 64-bit skype (?).
Not sure.
Virtual Windows on Linux
While looking for options to load my Garmin from Ubuntu, I came accross a link to Windows-7 Release Candidate. The activation looks to be good for about a year.
I have experience with both VMWare and VirtualBox hosting Linux on Windows (Android development). I've always wanted to give this a try the other way around (but wasn't eager for the cost of another Windows license).
A Windows-7 Customer Preview of the release candidate is available this month, so I gave it a try. The good news is it's to be active thru May of 2010.
I got VirtualBox 3.0.0 first for Ubuntu 9.04, following the instructions for adding it to the sources list. I added the source list, apt-get update and then searched for the virtualbox-3 version the site spoke of.
My Ubuntu installation is a 32 bit (even though my hardware is 64 bit), so I installed the 32-bit version of VirtualBox, as well as a 32 bit version of Windows-7. After installing VirtualBox, a reboot showed it in the Applications->System Tools menu.
Registration worked, I made a new Windows-7 "disk", allocating 1024M RAM to it, and then added the Windows-7 installation CD iso image to this "machine." I was now ready to "boot" this virtual instance. Windows installation worked fine. Next step was to reboot and update windows. On my setup (at least) the cursor-capture (right control) key doesn't seem to get passed back to VirtualBox once Windows-7 is running (it will before windows boot has completed, it seems), so a priority was to install the Virtual Box Guest additions from the Devices Menu. Installing this CD image showed it mounted on Windows (from the Start menu, Computer). I ran theVboxGuestAdditions-x86 version, rebooted Windows and now my mouse was no longer captive.
UPDATE: Note that for the VBoxGuestAdditions to correctly install on Windows-7, you need to boot Windows-7 in "safe" mode: press F8 at boot to get a menu giving you the option. Otherwise, Windows will not allow some of the files to be installed.
Browser, network, updates, sound, etc. all worked.
On Fedora-11 x86_64, again things are a little more complicated.
From www.virtualbox.org I installed the Fedora-11 64 bit version (an RPM file), and downloaded the 64 bit version of Windows-7 Preview (the release candidate). Reboot of Fedora-11, and VirtualBox appeared on my Applications->System Tools menu, but wouldn't boot (critical error). Firing it up from a shell (VirtualBox) showed a permissions problem on startup. On Fedora-11 the first userid is 500; on Ubuntu it's 1000. Since I have a dual-boot setup, and since I share the source (and document) areas, I changed my userid under Fedora. VirtualBox makes an IO directory, /tmp/.vbox - BUT incredibly it does not create it with the current invoking user's ID, but with (!) 500. Chown (with group = vboxuser) of this directory solved my startup problem (good). Unfortunately I'll have to do this each time I boot Fedora (it's tmp! I've reported this). Once running, installing Windows-7 preview worked fine; I followed the same procedures as for Ubuntu. On the Fedora Vbox, I could not get sound to work - whatever combination of vbox settings I used, Windows-7 did not see an audio device under the 64-bit setup & Fedora.
Other than that, browsing, windows update, playing videos (youtube, etc.), songs, all the fundamentals seem solid. On the 64 bit side, Adobe's Flash player didn't seem to be able to load automatically in Win-7 - I had to manually select a 64 bit windows (Vista) option for the browser install, and then it worked. Performance on both seems pretty darn good.
Another way to test web applications. I've built Google Chrome for Linux, but it's still in Alpha. Now I have Chrome, Firefix, Safari, IE8, and Opera running under windows.
As I fire these up, and try to make some links, I'm seeing the control key in VirtualBox 3 doesn't seem to come thru to the applications in the guest, e.g. CTRL-C / CTRL-V to copy / paste don't seem to work at all (I have to right-click COPY / right-click PASTE). Checking this with copy/paste attempt from within wordpad confirms; only the 'c' of a CTRL-c keypress makes it through; menu pasting pastes something from the paste buffer from before (outside windows).
I have experience with both VMWare and VirtualBox hosting Linux on Windows (Android development). I've always wanted to give this a try the other way around (but wasn't eager for the cost of another Windows license).
A Windows-7 Customer Preview of the release candidate is available this month, so I gave it a try. The good news is it's to be active thru May of 2010.
I got VirtualBox 3.0.0 first for Ubuntu 9.04, following the instructions for adding it to the sources list. I added the source list, apt-get update and then searched for the virtualbox-3 version the site spoke of.
My Ubuntu installation is a 32 bit (even though my hardware is 64 bit), so I installed the 32-bit version of VirtualBox, as well as a 32 bit version of Windows-7. After installing VirtualBox, a reboot showed it in the Applications->System Tools menu.
Registration worked, I made a new Windows-7 "disk", allocating 1024M RAM to it, and then added the Windows-7 installation CD iso image to this "machine." I was now ready to "boot" this virtual instance. Windows installation worked fine. Next step was to reboot and update windows. On my setup (at least) the cursor-capture (right control) key doesn't seem to get passed back to VirtualBox once Windows-7 is running (it will before windows boot has completed, it seems), so a priority was to install the Virtual Box Guest additions from the Devices Menu. Installing this CD image showed it mounted on Windows (from the Start menu, Computer). I ran theVboxGuestAdditions-x86 version, rebooted Windows and now my mouse was no longer captive.
UPDATE: Note that for the VBoxGuestAdditions to correctly install on Windows-7, you need to boot Windows-7 in "safe" mode: press F8 at boot to get a menu giving you the option. Otherwise, Windows will not allow some of the files to be installed.
Browser, network, updates, sound, etc. all worked.
On Fedora-11 x86_64, again things are a little more complicated.
From www.virtualbox.org I installed the Fedora-11 64 bit version (an RPM file), and downloaded the 64 bit version of Windows-7 Preview (the release candidate). Reboot of Fedora-11, and VirtualBox appeared on my Applications->System Tools menu, but wouldn't boot (critical error). Firing it up from a shell (VirtualBox) showed a permissions problem on startup. On Fedora-11 the first userid is 500; on Ubuntu it's 1000. Since I have a dual-boot setup, and since I share the source (and document) areas, I changed my userid under Fedora. VirtualBox makes an IO directory, /tmp/.vbox - BUT incredibly it does not create it with the current invoking user's ID, but with (!) 500. Chown (with group = vboxuser) of this directory solved my startup problem (good). Unfortunately I'll have to do this each time I boot Fedora (it's tmp! I've reported this). Once running, installing Windows-7 preview worked fine; I followed the same procedures as for Ubuntu. On the Fedora Vbox, I could not get sound to work - whatever combination of vbox settings I used, Windows-7 did not see an audio device under the 64-bit setup & Fedora.
Other than that, browsing, windows update, playing videos (youtube, etc.), songs, all the fundamentals seem solid. On the 64 bit side, Adobe's Flash player didn't seem to be able to load automatically in Win-7 - I had to manually select a 64 bit windows (Vista) option for the browser install, and then it worked. Performance on both seems pretty darn good.
Another way to test web applications. I've built Google Chrome for Linux, but it's still in Alpha. Now I have Chrome, Firefix, Safari, IE8, and Opera running under windows.
As I fire these up, and try to make some links, I'm seeing the control key in VirtualBox 3 doesn't seem to come thru to the applications in the guest, e.g. CTRL-C / CTRL-V to copy / paste don't seem to work at all (I have to right-click COPY / right-click PASTE). Checking this with copy/paste attempt from within wordpad confirms; only the 'c' of a CTRL-c keypress makes it through; menu pasting pastes something from the paste buffer from before (outside windows).
Pidgin for online communications...
In my corporate environment, I had a couple of choices to cross over all the various messaging and internal, secure chat rooms, as well as outside connection for occaisional things like PyCon IRC meetings.
I used several things, eventually settling, because of breadth of coverage in one place, on pidgin.
Fedora-11 seems to keep up-to-date with pidgin releases (it just updated me this past weekend), but Ubuntu does not - so I followed the instructions for adding the repository, as shown on http://www.pidgin.im/download/ubuntu/
So far, so good - my Ubuntu now stays up-to-date with pidgin, albeit straight from the pidgin sources. If I have problems, I'll have to post bugs directly to pidgin.
I used several things, eventually settling, because of breadth of coverage in one place, on pidgin.
Fedora-11 seems to keep up-to-date with pidgin releases (it just updated me this past weekend), but Ubuntu does not - so I followed the instructions for adding the repository, as shown on http://www.pidgin.im/download/ubuntu/
So far, so good - my Ubuntu now stays up-to-date with pidgin, albeit straight from the pidgin sources. If I have problems, I'll have to post bugs directly to pidgin.
nslookup?
For some reason, nslookup was not working on Fedora-11 today...
Some of the symlinks were illformed for the libraries.
Some of the symlinks were illformed for the libraries.
$sudo yum reinstall bind-libsseemed to solve the problem. The re-install was about 1/2 the size of the uninstall... not sure what was up.
Fedora Browsers
Blogger dot com complained of an error, post failing when I tried posting from IE6 under wine; apparently that post happened anyway.
As follow-up, when I tried to install IEs4Linux using it's gui, wine failed on rundll - wine complaining:
This is with wine 1.1.23 (from Fedora).
The first time through, the gui installer caused Gtk to complain - something about some code in a .c file that should never be reached... I could not get this to repeat, so it must be something about file creation / tests / etc.
I ran the installer with no-qui, and was able to get IE6 working. I tried making this post with it (and in fact had completed the post content) - but when trying to make the post, blogger encountered a submission error, and IE6 didn't keep a cache - turns out the submission happened anyway.
So, I went to FIrefox to redo the post - only to find it there.
Not sure how useful testing with this kind of IE will be - it looks like basic browsing and rendering are useful.
The beta IE7 install fails to reach anything; I've given up on it. For grins, I installed the IE5.5 (not sure what I'd want that for).
Perhaps I'll try this over on my ubuntu, where - if memory serves me - I have wine 1.1.24 running (and it's a 32 bit environment).
As follow-up, when I tried to install IEs4Linux using it's gui, wine failed on rundll - wine complaining:
This is with wine 1.1.23 (from Fedora).
The first time through, the gui installer caused Gtk to complain - something about some code in a .c file that should never be reached... I could not get this to repeat, so it must be something about file creation / tests / etc.
I ran the installer with no-qui, and was able to get IE6 working. I tried making this post with it (and in fact had completed the post content) - but when trying to make the post, blogger encountered a submission error, and IE6 didn't keep a cache - turns out the submission happened anyway.
So, I went to FIrefox to redo the post - only to find it there.
Not sure how useful testing with this kind of IE will be - it looks like basic browsing and rendering are useful.
The beta IE7 install fails to reach anything; I've given up on it. For grins, I installed the IE5.5 (not sure what I'd want that for).
Perhaps I'll try this over on my ubuntu, where - if memory serves me - I have wine 1.1.24 running (and it's a 32 bit environment).
Browser News
Fedora-11 update today installed a Firefox 3.5-1
Neat stuff; some new things, among them custom buttons.
As part of custom buttons, I started looking through what's available (thinking I'd get a gmail button, for example).
One of the things I saw was IE6 for Linux. Interesting. Internet Explorer launched through something called IEs4Linux.
I searched - target audience: "Web designers who are on Linux"....
It seems to use a wine, configured so that it's a separate instance - so it doesn't mess up your registry on your "main" wine.
I tried installing it - it failed in gui mode (Gtk complained about code that should never be reached!). The installation from the command line seemed to have a problem when installing flash-9 (something about a rundll failing), but when I tried the browser, all seemed operational. I ran over to google.com (which tried to get me to install chrome - sorry, linux native version, thank you... more about that later). I went to www.web2py.com which now has flash-driven slides on the front page - and IE6 wanted me to install flash-player... ok, let's try, see if _this_ has a rundll failure... Nope; no problem; I'm watching flash content in my IE6 on linux.
This post is written from an IE6, running on wine 1.1.23 on Fedora-11.
Fancy that!
-----------------
Postscript - Apparently, despite blogger complaining, this did post from IE6
Neat stuff; some new things, among them custom buttons.
As part of custom buttons, I started looking through what's available (thinking I'd get a gmail button, for example).
One of the things I saw was IE6 for Linux. Interesting. Internet Explorer launched through something called IEs4Linux.
I searched - target audience: "Web designers who are on Linux"....
It seems to use a wine, configured so that it's a separate instance - so it doesn't mess up your registry on your "main" wine.
I tried installing it - it failed in gui mode (Gtk complained about code that should never be reached!). The installation from the command line seemed to have a problem when installing flash-9 (something about a rundll failing), but when I tried the browser, all seemed operational. I ran over to google.com (which tried to get me to install chrome - sorry, linux native version, thank you... more about that later). I went to www.web2py.com which now has flash-driven slides on the front page - and IE6 wanted me to install flash-player... ok, let's try, see if _this_ has a rundll failure... Nope; no problem; I'm watching flash content in my IE6 on linux.
This post is written from an IE6, running on wine 1.1.23 on Fedora-11.
Fancy that!
-----------------
Postscript - Apparently, despite blogger complaining, this did post from IE6
User Space Across Fedora & Ubuntu
I started to boot in different operating systems depending on what I was doing.
For example I have Bruce Eckel's First Steps in Flex book - but I've read about Adobe's Flex Builder for Linux just installing and working better in 32-but Linices, so I do that work in Ubuntu... but I review and do some code work when in Fedora.
Another example is web2py which I update from launchpad and do tests on (intentionally) in both.
Same goes for finances, Tomboy notes, web links, gnucash files, documents.... there are just many things I want to have one common copy for regardless of where I boot.
Ubuntu starts the default first userID at 1000. Fedora starts the first default userID at 500.
I created a second user on each (handy regardless, for backing up - I use rsync - the home space of my "main" account, since I don't want to be logged in when doing that).
On fedora, I changed the user and group id of my "main" user account in /etc/passwd and /etc/groups, then recursively changed the owner of the home space (now of an unknown user):
Recall that the common denominator is Ubuntu, since it uses ext3 filesystesm, while Fedora-11 now defaults to ext4 (for all except the /boot partition).
By the way, the internal disk seems quieter (an old EIDE drive) and noticeably faster on ext4 activities than ext3 (when those are on the same drive), so I like / prefer using ext4 for now.
I have not yet symlinked Tomboy - not sure if any of the data is dependent on 32-bit or 64-but aspects yet.
For example I have Bruce Eckel's First Steps in Flex book - but I've read about Adobe's Flex Builder for Linux just installing and working better in 32-but Linices, so I do that work in Ubuntu... but I review and do some code work when in Fedora.
Another example is web2py which I update from launchpad and do tests on (intentionally) in both.
Same goes for finances, Tomboy notes, web links, gnucash files, documents.... there are just many things I want to have one common copy for regardless of where I boot.
Ubuntu starts the default first userID at 1000. Fedora starts the first default userID at 500.
I created a second user on each (handy regardless, for backing up - I use rsync - the home space of my "main" account, since I don't want to be logged in when doing that).
On fedora, I changed the user and group id of my "main" user account in /etc/passwd and /etc/groups, then recursively changed the owner of the home space (now of an unknown user):
$ sudo chown -R myid:mygroup /home/myidNow, key areas (Downloads, Documents, Pictures, Music, projects, workspace....) I symlinked to my Ubuntu space.
Recall that the common denominator is Ubuntu, since it uses ext3 filesystesm, while Fedora-11 now defaults to ext4 (for all except the /boot partition).
By the way, the internal disk seems quieter (an old EIDE drive) and noticeably faster on ext4 activities than ext3 (when those are on the same drive), so I like / prefer using ext4 for now.
I have not yet symlinked Tomboy - not sure if any of the data is dependent on 32-bit or 64-but aspects yet.
Getting the Epson CX7400 scanner working (Fedora & Ubuntu)
Printing has worked well in both Fedora & Ubuntu, but SANE didn't recognize my scanner.
This works in Ubuntu (I expect it will in F11; will add a note when I try). Looking at the sane-project site, I edited /etc/sane.d/epson2.conf, adding:
to list scan devices. This didn't work, but it suggested I try
This found the scanner, but reported that a back-end might be needed.
A little googling got me to try
Installing the iscan_2.20.0-6.ltdl7_i386.deb driver from http://www.avasys.jp/english/linux_e/ did the trick.
I did some test scans, and found at least one bug:
It turned out to be more problematic to get the scanner working on Fedora-11. The scanner id (sane-find-scanner) was already in place (not sure - perhaps because of previous action of mine?), so I got the iscan-2.20.0-6.x86_64.rpm driver from http://www.avasys.jp/english/linux_e/. It wouldn't install, complaining of a missing /usr/lib64/libltld.so.3 version of a libtools library. There was a version 7, so I tried linking ...so.3 to point to it - no good. I looked through the avasys forums, and found a note about Ubuntu and missing dependency so I decided to try something similar for Fedora.
I searched for a Fedora package that contained this libtools version of the library: I found libtool-ltdl-1.5.26-4.fc10.x86_64.rpm and downloaded it locally. Yum wouldn't install it ("It doesn't update anything." - which it doesn't; I already have a newer package installed).
Someone versant in yum may know how to install older packages when you need them too. I did this:
This works in Ubuntu (I expect it will in F11; will add a note when I try). Looking at the sane-project site, I edited /etc/sane.d/epson2.conf, adding:# for Epson CX7400 scanner, I'm trying this:I still didn't get a scanner showing up in gimp. Based on help from gimp, I tried
usb 0x4b8 0x0838
$ scanimage -L
to list scan devices. This didn't work, but it suggested I try
$ sane-find-scanner
This found the scanner, but reported that a back-end might be needed.
A little googling got me to try
$sudo apt-get install libsane-extrasThis still didn't help.
Installing the iscan_2.20.0-6.ltdl7_i386.deb driver from http://www.avasys.jp/english/linux_e/ did the trick.
I did some test scans, and found at least one bug:
- I acquired a preview from the interface. The default scan settings showed 300 dpi (x and y), but the scan got only the preview resolution.
- Changing the dpi settings to 600x600 showed something dramatically larger;
- changing it back to 300x300 this time got a 300x300 image scan.
It turned out to be more problematic to get the scanner working on Fedora-11. The scanner id (sane-find-scanner) was already in place (not sure - perhaps because of previous action of mine?), so I got the iscan-2.20.0-6.x86_64.rpm driver from http://www.avasys.jp/english/linux_e/. It wouldn't install, complaining of a missing /usr/lib64/libltld.so.3 version of a libtools library. There was a version 7, so I tried linking ...so.3 to point to it - no good. I looked through the avasys forums, and found a note about Ubuntu and missing dependency so I decided to try something similar for Fedora.I searched for a Fedora package that contained this libtools version of the library: I found libtool-ltdl-1.5.26-4.fc10.x86_64.rpm and downloaded it locally. Yum wouldn't install it ("It doesn't update anything." - which it doesn't; I already have a newer package installed).
Someone versant in yum may know how to install older packages when you need them too. I did this:
$ sudo rpm -i --oldpackage libtool-ltdl-1.5.26-4.fc10.x86_64.rpmTrying the libtool command, its version seems unaffected (and loading the right thing). Now installing the iscan 64 bit version succeeded, and Gimp scans with Xsane (same caveat about inital resolution settings not taking applies). Iscan doesn't have this problem (available from gimp in both Ubuntu & Fedora).
$ ls /usr/lib64/liblt*
/usr/lib64/libltdl.so.3 /usr/lib64/libltdl.so.7
/usr/lib64/libltdl.so.3.1.6 /usr/lib64/libltdl.so.7.2.0
$
Fedora Update Breaks Boot (NVIDIA driver)
I was a little miffed when this happened. I solved it. Here's the background.
First, the NVIDIA driver installation (which is just marvellous) didn't find pre-compiled kernel interfaces, and so as part of it's installation compiled headers and generated kernel interfaces to the driver.
This broke with a kernel update.
After the F-11 update, boot would come up, and when X started (I was expecting login dialogue), I got a blank, dark screen, no cursor. Unfortunately, Fedora 11 does NOT come with the nice boot screen that Ubuntu does. Fortunately, you can change it. (More on that later).
So no amount of reboot attempts got me anywhere, so I tried Ubuntu - to get to alternate boot device from the ASUS BIOS takes a ALT-F8 key-press at start-up. Paranoid, I picked an Ubuntu "recovery" mode (I have no idea what that is, does, never tried it before). It checked disks, scanned hardware (if I remember correctly) and then rebooted - this time I went into normal boot and it came up.
Phew!
Since NVIDIA was running on Ubuntu (I had dual monitors again) I guessed this must be what happened - new kernel. To test, I wanted to boot to console - init 3 instead of init 5 for Fedora.
'mount' showed me my ubuntu usb was on /dev/sdb1; /dev/sda was my internal drive. Beginning with F-11, Fedora created filesystems with ext4, and ubuntu doesn't support this. I booted from the Fedora live CD and mounted the fedora disk.
I looked at the grub menu from Ubuntu, and compared it with the grub menu in fedora - buggers! I enabled a 10 second timeout in Fedora, and enabled the menus (fedora's grub was still allowing booting in recovery mode, or in past kernels).
I strongly encourage changing this in your Fedora-11:
While this is annoying, at least it's simple and relatively quick.
I'm not sure - and will watch - if a similar kernel update in Ubuntu will have the same effect (I don't think so, as I've had kernel updates, always used the NVIDIA driver, and never had to do this manually).
Finally, I thought it would be nice to have the ubuntu usb on my Fedora Grub menu. There are a couple of things different between Fedora and Ubuntu grubs (even though they both report grub 0.97):
First, the NVIDIA driver installation (which is just marvellous) didn't find pre-compiled kernel interfaces, and so as part of it's installation compiled headers and generated kernel interfaces to the driver.
This broke with a kernel update.
After the F-11 update, boot would come up, and when X started (I was expecting login dialogue), I got a blank, dark screen, no cursor. Unfortunately, Fedora 11 does NOT come with the nice boot screen that Ubuntu does. Fortunately, you can change it. (More on that later).
So no amount of reboot attempts got me anywhere, so I tried Ubuntu - to get to alternate boot device from the ASUS BIOS takes a ALT-F8 key-press at start-up. Paranoid, I picked an Ubuntu "recovery" mode (I have no idea what that is, does, never tried it before). It checked disks, scanned hardware (if I remember correctly) and then rebooted - this time I went into normal boot and it came up.
Phew!
Since NVIDIA was running on Ubuntu (I had dual monitors again) I guessed this must be what happened - new kernel. To test, I wanted to boot to console - init 3 instead of init 5 for Fedora.
'mount' showed me my ubuntu usb was on /dev/sdb1; /dev/sda was my internal drive. Beginning with F-11, Fedora created filesystems with ext4, and ubuntu doesn't support this. I booted from the Fedora live CD and mounted the fedora disk.
I looked at the grub menu from Ubuntu, and compared it with the grub menu in fedora - buggers! I enabled a 10 second timeout in Fedora, and enabled the menus (fedora's grub was still allowing booting in recovery mode, or in past kernels).
I strongly encourage changing this in your Fedora-11:
- edit /boot/grub/grub.conf;
- change timeout;
- comment out hiddenmenu:
# timeout=0
timeout=10
splashimage=(hd0,0)/grub/splash.xpm.gz
# hiddenmenu
While this is annoying, at least it's simple and relatively quick.
I'm not sure - and will watch - if a similar kernel update in Ubuntu will have the same effect (I don't think so, as I've had kernel updates, always used the NVIDIA driver, and never had to do this manually).
Finally, I thought it would be nice to have the ubuntu usb on my Fedora Grub menu. There are a couple of things different between Fedora and Ubuntu grubs (even though they both report grub 0.97):
- Ubuntu boot menu is in /boot/grub/menu.lst; Fedora is in /boot/grub/grub.conf
- Ubuntu (nicely) identifies disks by UUID; Fedora grub does not appear capable of this;
title Ubuntu 9.04, kernel 2.6.28-13-genericFedora's grub doesn't like this; I changed the uuid line to:
uuid 2........02da
kernel /boot/vmlinuz-2.6.28-13-generic root=UUID=2.........02da ro quiet splash
initrd /boot/initrd.img-2.6.28-13-generic
root (hd1,0)and this seems to make everything happy (the uuid on the kernel line remains intact) - my other (backup drives) are off of a usb hub, so it seems this usb is always showing as hd1, which is good.
Kickoff
I thought I'd document progress on my new desktop adventures.
I'm becoming a Linux-only home office.
I thought about what I _really_ didn't want to live without from my PC, and if there were equivalents. In sum, I found replacements. The one or two things that I didn't know how to replace I got to run under wine. Good-bye traditional PC!
What I'm running on:
Asus M3N78 Pro; AMD Phenom 9500;
Those are the basics. I had some trouble bringing up these boards and getting stable running, but here are some basics of what I ended with:
Fedora-10 (what I'd used on a 64 bit machine at work) didn't work any better (worse, if I recall). I tried Fedora-11 preview, and it seemed to work all day long; games, internet, devices, mounting, etc. Two days later, Fedora-11 was released and I loaded that.
I posted a report to Ubuntu forums about my problems, and was encouraged to try 8.04 or 8.10 Ubuntus to see what my experience was. I wanted to get a working system, but finally did try 8.04 x86_64, and it seemed to work well. By now, I was convinced there was nothing wrong with the mother board / processor.
Later, I would try to boot from by Ubuntu-32 bit 9.04 usb installation (I don't know why I didn't try this sooner) - I tried the "recovery" mode first, and that seemed to test and reconfigure (?) things - in any case, booted and worked well - except for the NVIDIA driver - it would lock on some gnome games (completely - no mouse, no keyboard; I had to powerdown). Upgrading to the NVIDIA driver (see below) solved this problem. I am writing this from my Ubuntu-32 9.04 boot from the usb. Great stuff.
I wish I had documented more of this as I was going, so I will continue to.
I'm becoming a Linux-only home office.
I thought about what I _really_ didn't want to live without from my PC, and if there were equivalents. In sum, I found replacements. The one or two things that I didn't know how to replace I got to run under wine. Good-bye traditional PC!
What I'm running on:
Asus M3N78 Pro; AMD Phenom 9500;
Those are the basics. I had some trouble bringing up these boards and getting stable running, but here are some basics of what I ended with:
- Ugrade ASUS BIOS to 1003;
- Find an O/S which seems to run stably on this board;
Fedora-10 (what I'd used on a 64 bit machine at work) didn't work any better (worse, if I recall). I tried Fedora-11 preview, and it seemed to work all day long; games, internet, devices, mounting, etc. Two days later, Fedora-11 was released and I loaded that.
I posted a report to Ubuntu forums about my problems, and was encouraged to try 8.04 or 8.10 Ubuntus to see what my experience was. I wanted to get a working system, but finally did try 8.04 x86_64, and it seemed to work well. By now, I was convinced there was nothing wrong with the mother board / processor.
Later, I would try to boot from by Ubuntu-32 bit 9.04 usb installation (I don't know why I didn't try this sooner) - I tried the "recovery" mode first, and that seemed to test and reconfigure (?) things - in any case, booted and worked well - except for the NVIDIA driver - it would lock on some gnome games (completely - no mouse, no keyboard; I had to powerdown). Upgrading to the NVIDIA driver (see below) solved this problem. I am writing this from my Ubuntu-32 9.04 boot from the usb. Great stuff.
- Install NVIDIA driver for GeForce 8300 (version 185.18.14)
I wish I had documented more of this as I was going, so I will continue to.