Unable to start xserver on Mandrake 9.2
Hey, I'm basically a linux noobie, and have a problem with one of my machines I use for work.
As the title states I am unable to start my xserver, using the startx command I have been using for the past 5 years. It goes to the blue loading screen with the x for a cursor, hangs there for about a min or so then goes back to the command screen. I get an error: .. .. giving up. xinit: Permission denied (errno 13): unable to connect to xserver waiting for xserver to shut down xinit: Server error. I have tried logging on as root and trying startx again. Same error. Any help to diagnose/solve this problem would be great. |
Mandrake 9.2 is from 2003. It is unsupported for a very long time, you don't get any security updates and bugfixes and haven't got them for the past years. If that machine is connected to the net it is wide open for everyone to break into that system. Why are you using such an old version?
|
The machine is on a closed network, and connected to a printer via scsi for the past 10 years.
The printer is still very useful, but the software that drives the printer is very old, and I believe can only run on a mandrake environment. I understand finding help for this old version is probably slim to none. I wanted to try everything before I attempt a fresh install, as I know very little, and may not ever be able to get back running again. Since my first post. I tried using the install CD's to find a repair utility. No dice, all I can find is an upgrade to 10.x or install fresh. That may be my only option. |
Your hidden .Xauthority file might be the culprit. Copy it to another location, a new one will be created automatically. See if that helps.
Before attempting a fresh install, try using a live cd and see if your printer is still supported. |
Thanks for the idea, Nexus.
I moved the file, and then tried running startx again. Stalled the same way. When it returned to the command screen I checked that it made a new file with the ls -a command. Sure enough it made a new one. The printer I use is a large format machine and can only be driven using this old software. As far as I know there is no standard printer driver for it. So a live CD would probably not help unless I could load this rip software on it as well. Any other ideas? |
What model printer? What software are you using with it? Do you know what model video card is in the machine? If you don't know please post the output of
Code:
lspci Lets see if we can't find where things are going wrong before suggesting further solutions. |
It's an x2Tech, from when Xerox dabbled in color graphics ~9 years ago. To this day it still prints faster than most printers on the market. The rip software I use is Caldera, a french imaging software.
I'm sorry. The lspci command posted a lot of info. Not sure if you need it all, but the bottom return is nVidia GeForce4 Ti 4200 I have no Xorg.0.log, or xorg.conf file. In either of those directories. Closest to that name is XF86Config in /etc/X11/ or XFree86.0.log in /var/log If that is what you want posted to pastebin let me know. I unsure how to get the contents of those files to a text file on a usb, but I'll try and find that tonight while I'm at home. |
The good news is Caldera RIP (from version 5.06 up to version 7.22) will work with Mandriva 2005. The discs can be downloaded from DistroWatch.
Newer versions of Caldera RIP run on other distros see here: Compatible Distributions. That's all the info I need with regards to the make/model of graphics card. The contents of XF86Config and XFree86.0.log will do. You can get them into a text file by typing (no need to navigate to the actual directories): Code:
cat /etc/X11/XF86Config > xfconfig.txt Code:
cat /var/log/XFree86.0.log > xfree86.txt |
I made the txt files, but am having trouble mounting and getting the files to a usb drive...
They are too large for a floppy. |
Compress them:
Code:
gzip -9 NAMEOFTEXTFILE |
xfree86.txt link: http://pastebin.com/8MTrqNNB
xfconfig: Code:
Section "Screen" |
Nothing flagged up as an error in your xfree86 log. I can see something stange about your xfconfig though. The Driver section should have nv or nvidia not svga.
You can generate a new xfconfig by running as root: Code:
XFdrake Another option would be to select vesa as the vidoecard option, that should at least allow you to log in to a desktop. Either way you'll get the opportunity to test your selection to see if the X server will work. |
I chose the GeForce 4 (generic) from the list, went to test and the screen went black and never came back. I had to hard reboot after 10min.
I just tried the vesa option under xfree4, same deal, I am looking at a black screen. |
Also I wanted to thank you for taking the time, and I hope this is not as frustrating on your end as it is on mine.
|
I can sympathise with your frustration. I'm happy to help out when time permits :) Can you take another look at your xfconfig and tell me what it says under the Driver section. Just use:
Code:
cat /etc/X11/XF86Config |
Ok, I got the report but it zoomed by really fast.
Newbie question is there a way to make it go page by page instead of blasting the whole thing past me? |
Code:
| less Under driver it says "svga" |
Apologies I forgot it'll do that.
Code:
cat | less /etc/X11/XF86Config I have another thought, while we can spend a lot of time diagnosing the problem here's a quick test to see if it's related to your current login. If you have the permission to do so: 1. Create a new user with the command: Code:
adduser <username> Code:
passwd <username> |
logged in under new user. startx -> same blue screen with x for cursor -> after a min back to command prompt with same error.
|
How are you at editing via commandline? If you are able I'd like to see what happens if you manually change the svga in your XF86Config to vesa if that works and you get a desktop, you can use the Control Centre to change the video driver. If it fails, try changing it to nvidia and if that fails try nv.
But before doing any of that make a backup of your current XF86Config file to your current location with: Code:
cp /etc/X11/XF86Config . |
I meant the basic black command screen, not sure what is the proper term for that. The one it puts you in on booting
I am unsure on how to edit that file from that screen. Would I just change it in the XFdrake menu without using the test option? |
There's no special name for it. Terminal is generally used to let folks like me know you are at a command prompt. The XFdrake option doesn't seem to be generating a new XF86Config or we wouldn't still have svga under the driver section.
|
Yes, and no. It hung up when I when into the internal "Test" option it had, and had to hard re-boot.
So technically I never changed it, or so as I see it. Possibly force it to a different driver, without it's internal "test", then try and startx would change the XF86Config file? |
I see what you're saying, yes I agree do the changes but not the test and see if the changes get written to the XF86 file.
|
Ok, it definitely changed the format of that file when I chose the NVidia GeForce4 (generic) option.
But well down the page now it still says Driver "svga" even though it says Nvidia for vendor and board name above it. Also in XFdrake it said that the driver was supposed to be named nv... So it is properly formatting the file just not using the proper driver as I can tell. It's quitting time here again, though I am in tomorrow as well. If you are available I would love to resume this testing. Thanks again. |
Sure thing. I think we're looking at manually putting the name of the driver to use. I'm not sure if you have your own text editor installed. If not you'll need to read up on vi that will most certainly be there as the default commandline editor.
|
Ok, I worked in vi. Was pretty simple once I got the idea of command and input modes.
Changed the XF86Config file to read Driver "nv" like what it said in XFDrake. Checked it after I was back in the terminal, It did change to be named "nv". Tried a startx. Same hang, then same error. |
How about if you change it to nvidia? Also can you check if the nvidia module is being loaded, type:
Code:
lsmod | grep nvidia |
I just tried naming "nvidia" and same thing.
I ran the lsmod command and got lsmod:command not found. In root the lsmod command gave no return. |
I've just taken another look at the /etc/X11/XF86Config contents you posted. There seems to be a lot of info missing. I want to make sure we are working with the correct one. Please do a
Code:
ls /etc/X11 |
Yes there is. Here is a complete listing of the x11 directory.
Code:
app-defaults/ |
Can you post the contents of XF86Config-4 and XF86Config-4.old perhaps we can see what has changed.
|
XF86Config-4
Code:
# File generated by XFdrake. XF86Config-4.old Code:
# File generated by XFdrake. |
Quote:
Code:
cp /etc/X11/XF86Config-4 XF86Config-4.bkup First: Code:
telinit 3 Second: Code:
telinit 5 |
After doing all above. The screen goes black, 3 times, for a half second each time. Then I'm back in terminal.
After the telinit 5 command it says: Code:
INIT: Switching to runlevel: 5 |
Ooo Something interesting.
I decided to give startx a shot in my default user name. Returned with a new error. Code:
Waiting for X server to begin accepting connections (EE) failed to load module "nvidia" (module does not exist, 0) |
Ok time to take another look at what's being reported in your /var/log/XFree86.0.log can you do
Code:
cat /var/log/XFree86.0.log > xfree86.txt |
Can you change the driver section to say vesa Save the change, do the telinit 3 then telinit 5 then try to login as your normal user again.
|
I've got to step out. Let me know what the state of play is. I'll take a look at any problems when I get back.
|
I get the same blue screen where the "x" for the cursor is, but it is all distorted and there are 4 of them across the screen. Still no desktop or login option.
Here is a link to my XFree86.0.log: http://pastebin.com/QQfaPXeh |
No errors in your log. Back to XFdrake. Choose vesa and change your colours to 16 bit and your screen resolution to 640x480 do the test. If it appears to hang try Alt & Ctrl & Backspace to try and force an X server restart or try Alt & F2 - F6 to see if you can get a command prompt. Then do a graceful system restart with:
Code:
shutdown -r now |
No dice. I try the test option and get nothing but a black screen that I cannot get out of.
So I tried just setting it to vesa 640x480 and 16bit, shutting down orderly, then trying startx after the reset. Same thing. For fun I tried the nvidia 4 (generic) one as well @ 640x480 and 16bit, same thing. At any rate. It's time to hit the road for the weekend, have a drink, and resist the urge to punch little fluffy things. I'll be back here on Tuesday. Have a good weekend. |
When you get back let's create a new XF86Config file. Type:
Code:
XF86Config |
Ok gave that a shot. Created a new XF86Config file, after renaming the one I had. Signed in and tried startx. Same problem.
I was curious to see if it was the card itself, so I swapped in another GeForce card I had laying around, and got the same results. I also tried using the generic nvidia option instead of the geforce one, in the XF86Config file, but got the same results. Here is the contents of my current XF86Config file. (The one I generated with the xf86config dialog) Code:
# File generated by xf86config. |
Did you try using XFdrake again too? Go for vesa first, confirm the change has been written to XF86Config-4 before proceeding to startx. Do you know if you have your kernel sources on the machine? Take a look for files ending in .h in your usr/src/<your kernel version goes here>/include/linux. To get your current kernel version do:
Code:
uname -r |
Yesterday was crazy busy, sorry I did not have a chance to check in.
I just ran XFdrake and chose vesa, confirmed the change in my XF86Config-4 file. startx gave the same problem. Inside of the usr/src/ directory I only have another directory named RPM. Inside of that RPMS/, and inside of that are directories named athlon, i386, i486, i586, i686, k6 and noarch. Not sure if any of those are what I need to look into. My return from uname -r was: 2.4.22-10mkenterprise |
I don't see a directory called linux in what you've posted, so I don't think you have the kernel-source installed. Does this system have 1GB (or more) of ram? This seems to be the reason why the enterprise kernel was installed. I did some digging around and cannot find any kernel-source that specifically mentions enterprise. However there is this kernel-source: kernel-source-2.4.22-10mdk.i586.rpm I'm not sure if it can be used. The reason why I'm talking about kernel-source is after all that we've tried, it seems that the best way forward would be to do a manual re-install of the nvidia drivers. In order to do this you need to have the kernel-source already on the system that matches your current kernel version to compile against. You will also need to download the correct nvidia driver. If you decide to go down this route these are the files you need to download:
1. kernel-source-2.4.22-10mdk.i586.rpm available from here: http://rpm.pbone.net/index.php3/stat....i586.rpm.html (File Size: 173.66 MB) 2. GeForce4 Ti 4200 Driver Version 96.43.19 available here: http://www.nvidia.com/object/linux-d...19-driver.html (File Size: 15.7 MB) If you can get those two files and transfer them to the system, we can set about installing the kernel-source and then installing the nvidia driver. You only need to put them in a location where we can easily find them. |
I do have 2 512 chips in this system, not sure if that is why they chose the enterprise.
I have both of those files copied over to my root directory. I hope that kernel version works. Please advise on how to go about doing this. |
First we'll make sure both files are marked as executable:
Code:
chmod +x kernel-source-2.4.22-10mdk.i586.rpm NVIDIA-Linux-x86-96.43.19-pkg1.run To install the kernel-source type: Code:
rpm -ivh kernel-source-2.4.22-10mdk.i586.rpm 1. type: telinit 3 to kill the xserver (if you don't see a command prompt after hitting enter, press enter again) 2. type: ./NVIDIA-Linux-x86-96.43.19-pkg1.run the install process should now start. Follow the onscreen instructions. At some point you will be asked if you would like to download a module or kernel (I can't remember which) select no and continue with the install. 3. Once the install process has finished, no need to reboot, simply type: telinit 5 close your eyes, hit enter and hopefully when you open them again you have graphical login. Good Luck :) |
off of the rpm -ivh command, I got the following errors.
Code:
error: Failed dependencies |
All times are GMT -5. The time now is 06:04 AM. |