At a crossroads solving hardware issues

[cwaltrs]

I'd like to get everyone's opinions on what I should do next with my failing DIY Tivo project.

 

My original intent was to build a linux-based Tivo-like MythTv home system for around $250 as a fall/winter project. I'm now at a crossroads because the system isn't working and I'm not sure what to do next to fix my problems.

 

Here's the basic chronology:

 

1. Purchased a Dell gx110 with PIII 733Mhz, 512Mb RAM - $50

2. Installed a 250 Gb HD - $70

3. Installed Geforce2 w/s-video out - $30

4. Installed Happauge PVR-150 card - $90

5. Wireless ra2500 card - $10

 

So far, we're doing fine... $250 and a download of Mandriva 2006 and I'm rocking. Not so fast. The system has been very unstable - locking up and crashing suddenly. My first step was to replace an inadequate 200w power supply (I never considered what my power needs would be.) Cost $80 for a 360W dell-compatible, so now I'm up to $330 and the system is still unstable, in fact it seems worse. The motherboard looks clean, no popped capacitors, but I'm starting to suspect it nonetheless.

 

So what should I do now? I see 2 directions to go. I'd like your opinions...

 

1. Try to diagnose and fix what I have, even if it means more parts (e.g. motherboard)

2. Send back the power supply and look for another PC. Use all the hardware components I've already purchased.

 

My thoughts:

 

1. Try to diagnose and fix what I have.

Pros

This is likely to be the cheaper option because all the replaceable parts probably don't add up to $50. Plus, the PC is a joy to work with (roomy, modular and very expandable). I can (hopefully) get used parts like a motherboard on the web. It could be as simple as a failing cooling fan (not likely)

Cons

I don't really know *how* to diagnose the problem. I've been thinking about a strategy of swapping old and new parts to see what the heck might be the problem. Maybe there are tools on the Ultimate Boot CD that could help diagnose?

 

2.Look for a new PC

Pros

Make a new start with all I've learned. Send back the $80 power supply and use the money for a better PC (still over budget though)

Cons

Throwing away the $50 for the PC (no warranty). No guarantee that the problem isn't with the components (eg.hard drive) and not the PC itself. Need to get a PC that's compatible with the hardware I purchased.

 

So what do you think?

 

FYI, I've attached the details of my system config if you're interested.

 

Many thanks,

Charlie

gx110_hardware.txt

[arctic]

I think I would test the whole box with a Knoppix CD right now in order to see if it is only unstable when you use Mandriva or if the Live-cd also has some trouble. If Knoppix has problems, too, then I would try to find out the amount of power you really need and compare the power unit with the specs of your motherboard. Maybe the board/processor gets too little or too much power.

[jboy]

Also, I would try failsafe mode which will take you to the command line interface as the root user without network support. Is system stable? Then, if stable, you could type init 3 at the command prompt and go to run level 3 which would start network services. You could run the non-graphical lynx browser. Is system stable?

 

If system is stable in command line mode, maybe you've got a problem with your video setup. You could type XFdrake to re-configure video. You could start with a generic driver like VESA and go from there to try to pinpoint where the problem might be.

 

Also, you could run one of the memtest utilities that are available from the Ultimate Boot CD and probably also from the Knoppix LiveCD.

[cwaltrs]

All,

 

Thanks. I've actually been running the tops command all day in safe mode, so it looks like I'm on the right track.

 

BTW..

 

I noticed something about my hardware config that seems off. My (attached) report shows:

 

01:07.0 VGA compatible controller: nVidia Corporation NV18 [GeForce4

MX 4000 AGP 8x] (rev c1) (prog-if 00 [VGA])

Subsystem: XFX Pine Group Inc.: Unknown device 201f

Flags: bus master, 66Mhz, medium devsel, latency 64, IRQ 9

Memory at fd000000 (32-bit, non-prefetchable)

Memory at f0000000 (32-bit, prefetchable)

Expansion ROM at 80000000 [disabled]

Capabilities: [60] Power Management version 2

 

The board I installed is a XFX PVT98LQT Geforce MX4000 64MB 128-bit DDR PCI Video Card. I don't even have an AGP slot. Could this be it? I'll definitely run the video setup and verify everything.

 

Thanks

[AussieJohn]

That could be your problem, namely the video card is an AGP model but happens to fit in a pci slot ok. From what I understand, an AGP slot is similar to a pci slot but just dedicated to a video card. It is possible that it is close enough that it works but unstably without completely failing.

 

John.

[Ixthusdan]

The design of an agp card prohibits it from fitting into a pci slot. The contacts are further back than a pci, so it would be impossible to even force a fit. For Mandriva, pci functions used to be the standard. Occassionally, one had to add a line or two in order to assure that agp is fully functioning as it should. So, pci should be more stable, not less. Edit xconf for pci, not agp. However, a faulty southbridge could also be causing all of these symptoms particularly in light of a pci video card. Might be a motherboard problem. Would you list the symptoms, again, please?

[daniewicz]

Also, you could run one of the memtest utilities that are available from the Ultimate Boot CD

 

Yes. This would be my suggestion to test the RAM. Run memtest overnight.

[cwaltrs]

The design of an agp card prohibits it from fitting into a pci slot. The contacts are further back than a pci, so it would be impossible to even force a fit. For Mandriva, pci functions used to be the standard. Occassionally, one had to add a line or two in order to assure that agp is fully functioning as it should. So, pci should be more stable, not less. Edit xconf for pci, not agp.

 

I think I was on the wrong track here. The output of my lspci -v showed my video card with AGP in the description. The card and the slot are both PCI. I thought I might have selected an AGP driver by accident, which would have caused some wiggy behavior. However, I reinstalled the drivers, making sure it was a PCI driver, and the description still says AGP. I think its just an error in the text.

 

However, a faulty southbridge could also be causing all of these symptoms particularly in light of a pci video card. Might be a motherboard problem. Would you list the symptoms, again, please?

<{POST_SNAPBACK}>

 

My symptoms: the system simply will not go 24 hours without freezing up or crashing. Sometimes the system responds to Ctrl-Alt-F1. If not, I go through the Alt-SysReq series. The system only responds to the Alt-SysReq-b and reboots unhappily. Occasionally, the system will not even respond to the reboot command and I have to kill the power.

 

As you can see from my original post, I have done a lot to the PC, so it's hard to tell what might have happened. Here's my plan (following advice given in this post -- thanks):

 

1. Boot in failsafe mode and run for 24 hours with the top command running. I was already doing this when I started the post - PASSED

 

2. Run init 3 to initiate networking and run the system overnight, again running the top command to keep things active. I also played around with lynx, and it worked fine - PASSED. The system passed the test, but my ra0 did lose its wireless connection. At some point I need to figure this one out as well.

 

3. I'll run memtest overnight just to elimate that as a cause.

 

4. If all goes well, it would seem to indicate video as the culprit. I may try running with my old video card to see if the system works properly and stays running for 24 hours.

 

5. If that works, I'll reinstall my new video card and start with the default nvidia driver and work my way up to nvidia 4. FYI, the only reason I need a new video card is the s-video out. If there was a way to have VGA out convert to RCA composite or s-video , I'd skip the new card altogether.

 

That's the plan, which I'll adjust based on my progress and the kind insights from this crew.

 

Thanks,

Charlie

[cwaltrs]

Update...

 

I tried to restart my network services from the root prompt and the system hung trying to stop rao. I killed the process. I tried just doing service network start and the system hung again trying to load parameters. Finally I just typed reboot and the system was not able to reboot, it just hung there. I eventually just did the old Alt-SysReq-b, which worked. This sounds suspiciosly like the symptoms I've been seeing. Perhaps progress is being made and the wireless card is the culprit.

 

I think I'll go back to my regular ethernet connection and re-run my test plan. If all goes well, guess I'll have to learn the ndiswrapper wireless installation instead of using the native raO drivers. I'll also post in the software section and see if people have had similar trouble with ra0.

 

Thoughts?

Charlie