Well, there was an update today to the 2.6.35-2 kernel and all went well. smxi updated the kernel before the regular dist-upgrade and in the sequence I reinstalled the NVIDIA driver, rebooted and everything has been fine.
Back to top

I took a look a the sgfxi code, and all is as it should be.

default gets updated, grub.cfg gets updated, there should be no need to run update-grub.

Also, that variable is the one that gets updated.

My guess is some syntax change occured, but I'd have to see your grub.cfg and default file BEFORE the nvidia install to see what the problem was.

Other issues could be dkms being stuck in the initrd.img, but I can't say since only a small handful of people are reporting problems, and I'm getting no consistent data from them to debug.
Back to top

hi

Having seen more heat at sidux I finally swallowed the blue pill and am using damentz's kernel

smxi did have an issue with first install of it.....I could tell the module was not going too well.....but being a simple person I just exitted from the script and started again without reboot.

2) On reboot with dmz kernel I got the famous nouveau stuff....but smxi told me what to do....which was to reboot and run sgfxi
(from memory)

I don't have any logs to submit but thought I might mention it because if you do not read your screen with a major change....you might miss it?

3) The kernel removal....the advanced kernel options in smxi work very well for me.

I also ran....apt-get --remove purge sidux*....which finally separates my rude behaviour from the past.....and moves me into hyper gear into sid.

thanks again H2 for your script.

regards

gordy
Back to top

hi

today's DU 28 August 2010 has some duplicate downloads.

Its easier for me to show by picture

stashbox.org/987602/3.png

and not so easy for me to see in the log

stashbox.org/987604/smxi.log

The DU works .....it just downloads some packages again and again etc.

Not all were done this way.....I think maybe 10 packages at the most?
But by the time my brain cells functioned I had forgotten to do a screenshot.

regards
Back to top

if you are seeing busybox again and again that's something to worry about. Usually busybox stuff only appears if your actual root login failed and you got thrown into the utility repair shell.

But then again, the debian repo servers can and do screw up too, as do internet connections and various other things.

But whatever the case, this wouldn't be an smxi issue, it's either an apt or an internet issue.
Back to top

hi
EDIT

did a manual apt-get install test of software and issue is there,
so not smxi related

regards
Back to top

I've been having an unusual message for a while.



I used sysv-rc-conf to fix kdm runlevels manually but it didn't change a thing.

here's my [new user link]

I'm running aptosid upgraded from sidux
Back to top

I'm not sure, but it looks like you have a package with a bad post/pre install script, where exactly do you see that sed error?

If you can pinpoint the exact location I'll see if it's smxi, but my first guess is it's a bad package script in apt.
Back to top

Hi.
I installed here two kernels and then ran sgfxi -! 40. The script detected the kernels but exited saying "line <some number here> command not found".
Then I rebooted to one of the kernels, ran sgfxi, the script was updated, done the two stage installation (removed nouveau > reboot > install) and ran sgfxi -! 40 again just to see if it would work but again it gave the command not found error in line xxxx. So I installed to the other kernel with the option -K and that went well.

I don't have the exact number of the line because those failures to run sgfxi -! 40 didn't log anything. Actually after the error it ended with something like: "Did it work? Ok, Bye!"
Back to top

the failure doesn't seem to be in sgfxi, I have just tested that, and it all works as expected.

these kernel issues are almost ALWAYS caused by a borked apt install of a kernel package.

sgfxi -! 40 is tested and works as of now.

Do: aptitude install -f
or: apt-get install -f

If you can't grab a line number I can't really look for the error, all you have to do is run -! 40 in X, then copy the data, that doesn't need to be out of X.

It's clear to me you have a broken apt package, or that's my top guess, sgfxi is using some different things to build modules. different nvidia installer commands as well.

Assuming you have a newer nvidia card, 6xxx or newer, just do: sgfxi -cB
to install beta driver to current kernel only.

Start X, then run as root: sgfxi -B

this will let the installer run in your console, where you can easily read and copy any error messages. Without error messages neither you nor I can figure this out, but there is no error in the sgfxi logic that I can see for the -! 40 loop
Back to top