- main - earth.li

NVIDIA/ X
by Ricardo Campos 15 Feb '02

15 Feb '02

> Why would you want to play Q3 anyway? ;-) Rune and Tux Racer, actually! :-P, the driver provided by X is WELL below par (in fact unusable) for openGL games. HOWEVER, I now want it back, at least it damn well worked! :'( > No, the problem is that when they fry the system, you can't debug it. The > cause is that they are needlessly restricted. I get Mark's point, though. It would be a much better idea if NVIDIA left it open for the XFree86/kernel hackers to work on, cos they seem to have a better idea of what they're doing. Sorry to selfishly steer towards my problem. I still have a seriously fscked X configuration (i.e. it don't work), with no NVdriver module anywhere. So... this module- part of X or part of the kernel? Does the kernel load this module, which is provided by X, or vice versa? I think I'm getting confuZed here. Am I gonna have to install a kernel from source whether I like it or not now, to get the module back? OR X? How could I find the headers for my existing kernel, if they're not where they're meant to be, so I can rebuild the NVIDIA kernel patch from source? TIA Ricardo -- Get your free email from www.linuxmail.org Powered by Outblaze

2 2

Re: [Alug] recovery from disaster/rebuilding kernel
by MJ Ray 14 Feb '02

14 Feb '02

Adam wrote: > But the NVdriver *is* the proprietary Nvidia one [...] Yes, sorry, it seems that I was mislead by the OP saying they came with X. Why would you want to play Q3 anyway? ;-) > [...] the only problem is though that they are not open source =( No, the problem is that when they fry the system, you can't debug it. The cause is that they are needlessly restricted. -- MJR

1 0

Re: [Alug] Multiple NICs
by Ricardo Campos 14 Feb '02

14 Feb '02

> Well, you can use round-robin DNS which means that for the hostname you'll > assign two IP addresses. You just assign multiple A records for the > hostname. Sometimes the answer is so simple you just don't see it! Would this *just* provide fault tolerance (if one card goes down, it gives the next IP)? i.e. not load balance. Or does it alternate which IP it gives out to the client issuing the request? TIA Ricardo (PS Martyn- my RH system is now bang up-to-date- no probs whatsoever with the kernel upgrades!) -- Get your free email from www.linuxmail.org Powered by Outblaze

4 3

Re: [Alug] recovery from disaster/rebuilding kernel
by MJ Ray 14 Feb '02

14 Feb '02

Adam wrote: > No just build a new kernel, when you reboot your machine XDM and X will > not work, just grab the NVIDIA drivers from http://www.nvidia.com and make > sure you have them in /usr/src before you do all this. [...] Ahem, he said the NV driver that came with X, not the POS ones from nvidia.com. As long as AGP isn't disabled, it should work fine. I have an entirely healthy bias against the nvidia.com drivers from past experiences. -- MJR

2 1

Re: [Alug] recovery from disaster/rebuilding kernel
by Ricardo Campos 14 Feb '02

14 Feb '02

> No just build a new kernel, when you reboot your machine XDM and X will > not work, just grab the NVIDIA drivers from http://www.nvidia.com and make > sure you have them in /usr/src before you do all this. OK. So I decided to test this.... I used the 'which package to use' program from Nvidia. Backed up XF86Config and XF86Config-4. I downloaded the .rpms, and installed them as root in runlevel 4 (5 starts up X). I got an error with the kernel rpm: modprobe can't find NVdriver. find / | grep NVdriver can't find it either. Where did it go? :'( I rebooted to runlevel 4 again just incase. (5 starts x it's a RH7.2 machine). I run startx . Well, all I can say is... thank fsck for ext3! I fiddled with my XF86Config-4 file and managed to get it so it doesn't lock up, but it exits saying it can't find the NVdriver module still. So I downloaded the source tars. I run a make on the kernel modules. It can't find the headers! I boot into my trusty MDK8.0 system, get online, look up every FAQ I can find and some NVIDIA walkthroughs on linuxnewbie.I need to add (apparently) SYSINCLUDE=/lib/modules/2.4.9-21/build/include Only.... there's no such path. It stops at /lib/modules/2.4.9-21/ and the contents DOES have something called BUILD in it, that is flashing for some reason. I *did* recently update the kernel due to the myriad of security errata, but I made sure I downloaded the kernel-headers package too. erm.... *eep*! >all you have to do when you rebuild the > machine is make sure your new user account has the same UID no. or you can > use the chown command as root to get access back to the files when you > have rebuilt the machine. Of course! root always has a UID of 0! I can CHOWN the files to the new user acct. groovy. REMINDER TO SELF= Put a sticker on my monitor saying "If it ain't broke, don't fix it" -- Get your free email from www.linuxmail.org Powered by Outblaze

1 0

recovery from disaster/rebuilding kernel
by Ricardo Campos 14 Feb '02

14 Feb '02

I am going to attempt to rebuild my kernel (gulp). Now, if I TOTALLY screw it up (likely), I want to know if it is possible for me to preserve my /home partition, and my X configuration: 1. Will the new kernel build impact on my X configuration? I know that my kernel has the NVdriver module installed (the one that came with X 4.1.0), but I'm concerned as to whether if I rebuild the kernel from source, it will require a new installation of X? 2. I have a separate /home partition, on a separate drive (hdb) to where / resides (hda). I did this incase hda fails.However, as I understand it, even if I make a user with the same name and password and mount /home to the old /home partition, I will still not be able to access any of the files due to permissions- am I right? If so, would there be a way to save the user info on floppy/CD whatever so that if I did fsck it all up, I could just start an install from scratch and use those files to get access to the old /home? (an install from scratch being far quicker!) Could I possibly ask any more questions in one e-mail? ;) TIA Ricardo -- Get your free email from www.linuxmail.org Powered by Outblaze

3 2

Multiple NICs
by Ricardo Campos 14 Feb '02

14 Feb '02

OK, I'm wondering if this is possible in Linux: One server, thebeast.mydomain.com , has 2 NICS. 1 has an IP of 10.0.0.1, the other of 10.0.0.2 . It is connected to a switch (which has a few other servers attached), and the switch to the rest of the LAN. It's a pretty critical server, under quite heavy load. Is there any way that a client could connect to thebeast.mydomain.com and sometimes connect to one IP, sometimes to another? I'm looking for fault tolerance as well as load balancing between the network cards. I know that Intel Server adapters do this on NT and Novell. Anyone have any ideas? Would it involve DNS jiiggerypokery? i.e. DNS server sending requests to 1 IP then another. Or maybe a switch that could do it? Ricardo PS please don't say "buy another server and cluster it" or somesuch. an additional NIC is FAR cheaper than another high-end server. -- Get your free email from www.linuxmail.org Powered by Outblaze

2 1

RHN users-advice
by Ricardo Campos 12 Feb '02

12 Feb '02

Just a quick question for RedHatNetwork Users- anyone had any probs updating the kernel via rhnsd? I'm usiong the latest version of that... my only outdated packages are the kernel and XFree86.... (Yes, I'm scared it may fsck up, I only have a dial-up!) Ricardo _______________________________________________ "There are several codes, and I know several of them." -Mr. Precise. http://www.rscampos.net Check my online diary out: http://www.rscampos.net/blog.html -- Get your free email from www.linuxmail.org Powered by Outblaze

2 1

Re: [Alug] port forwarding
by Ricardo Campos 12 Feb '02

12 Feb '02

>If you firewall out SMTP into the network except for the > gateway then that should help you avoid doing a Demon and getting your > outgoing mail relay listed in an RBL even though it's actually a > customer's fault. What's an RBL? I use Demon as an ISP, and with a paid e-mail account I was only able to use demon's SMTP server, rather than the one I paid for... is this due to relaying? Would this also affect me if I had an MTA on my box at home? Ricardo -- Get your free email from www.linuxmail.org Powered by Outblaze

2 1

Re: [Alug] "Reverse" VNC
by Laurie Brown 12 Feb '02

12 Feb '02

Eric Sellin wrote: > Hi Laurie, > > If you want to gain control over a remote computer but you can't connect to > it (because of NAT, firewall or God knows what), you can start the VNC > viewer in Listen mode on your own computer. > > Then, on the remote computer, get your customer to:- > - start the VNC server > - right-click on the VNC icon in the icon tray > - choose the "Add New Client" option > - supply your own IP address > > Of course, there has to be somebody in front of the remote computer to > initiate the reverse connection, which I understand is not ideal. Thanks, Eric, that worked a treat. I'm a shade embarrassed I didn't spot that functionality myself! One thing I did have to do was to open port 550x (where x = display number) on my firewall to incoming traffic. Cheers, Laurie. -- --------------------------------------------------------------------- Laurie Brown laurie(a)brownowl.com PGP key at http://pgpkeys.mit.edu:11371 ---------------------------------------------------------------------

1 0