HELP! It's gone wrong, I tried to fix it and I'm making it worse.

[PhillJones]

I'm posting here because I think I have a hardware problem but I'm not sure so this is more your desperate cry for help than a specific question.

Since I don't know what is going wrong, the only thing I can do is tell the story so I apologize if this is a bit long and rambling.

My system is a dedicated HTPC running sage and very little else. My HDD setup is as follows

Drive 1:PATA (250 Gb)
C: System partition (30Gb)
D: Media: My music and photos (about 200Gb)

Drive 2: SATA (500Gb)
E: Sage recordings (300 Gb)
F: DVD rips of the kids favorite movies (about 250 Gb)

Problem 1:
about two weeks ago my HTPC started rebooting in the middle of the night and not restarting. There was an error that it couldn't find the system directory and that windows needed repairing. If rebooted it would work. I suspected it was overheating so opened the case, cleaned all the fans and heatsinks with an airduster. I take an image of the C drive just to be safe. Unplugged and replugged all the connections and it seems to work. I went away for a week and everything seemed fine.

Problem 2:
I get back and my wife says it's been working fine. A couple of days later sage crashes out, it just vanishes when we try and play any video. I wonder if something like a codec is corrupted but before I get a chance to try and figure it out, it starts doing the reboot and hang on restart thing again. I now suspect that my system drive is dying and this is making the seek time go through the roof and causing sage to crash.

Problem 3:
Last night I borrow a HDD, remove the existing system drive and put my backup onto the borrowed drive using ultimate boot CD. I neglect to make the second partition (Drive D. I then start the PC, everything seems to work but sage won't start, the hard drives just grind. It also won't let me get into drive management. I guess that sage can't find my music media directory and is looking endlessly for it since the correct drive doesn't exist. I restart using UBCD, wipe the borrowed drive and start again. This time I make the second partition and start the music files copying across. I then go to bed because this is going to take a while.

Problem 4:
This morning the PC is BSOD. I try to boot it up using my new borrowed system drive. Windows starts to load but I get the message that drive F may be corrupt, it checks drive F and it's fine but before the PC but immediately after that, before windows loads, the PC reboots. It now just cylcles through this process if I turn it on.

So basically, the more I try to fix this, the worse it gets and I don't know what to do next for fear of completely screwing the PC beyond repair.

Sorry, the post is long, but as you can tell, I don't know how to even describe the problem as I'm way out of my depth here. If you got this far, thanks for reading and if you reply, thanks again.

Phill

[Arioch5]

I would start by finding the manufacturer's test software for each hard drive. Typically you can burn a self booting CD that will allow you to test the drives, though you'll need one per manufacturer. Test each drive, you won't lose data, and you'll know if one (or multiple) of the drives is failing. Those test programs will also let you know if the drive is overheating (a problem I've had with one of my clients).

I'm just giving this my best guess, but windows will BSOD, hang, reboot, and do generally bad things when a HD is throwing to many errors. With you moving around a lot of files you're basically guaranteed to make windows puke if one of the drives has bad sectors.

[PhillJones]

Thanks Arioch.

I actually stuck my old system drive in my bag this morning and so have it with me. It's a seagate barracuda 300Gb PATA drive. I've found Seatools for windows and am running the long DST. It passed the short DST. I've also downloaded and burned the SeaTools CD so I can test both drives in the rig itself.

If you or anybody else have used barracuda drives and have run these tests. Can you confirm the the DST is better than the generic and if I run the Long DST file and it passes that it means that both the drive is fine and not overheating (obviously I intent to test both my original drives using the boot CD while they're installed in my HTPC rig.)

I nearly ran an advanced test but the programme told me that I could loose all my data if I did

[Article22]

It may or may not help but you can often get these kind of errors with memory, motherboard or cpu problems.

If a HD goes you generally know about it (grinding noises from HD, and nada ever again) which you haven't had...

I would take out any internal PC cards bar the graphics card (if you have one), set the bios to boot from cdrom first (here's hoping you have one) then boot the pc up with a Linux Live CD which has memtest on it, and run it for 2 hrs and see if it crashes, then you know whether it is the HD or not.

If it does crash then see if you have 2 sticks of memory, if so - remove one and re-test, then swap them over and re-test...if it stops crashing then you know it is the memory.

If it still crashes then your motherboard or cpu is probably fried, replace.

Hope you sort it out,

Nigel.

[PhillJones]

Originally Posted by Article22

It may or may not help but you can often get these kind of errors with memory, motherboard or cpu problems.

If a HD goes you generally know about it (grinding noises from HD, and nada ever again) which you haven't had...

I would take out any internal PC cards bar the graphics card (if you have one), set the bios to boot from cdrom first (here's hoping you have one) then boot the pc up with a Linux Live CD which has memtest on it, and run it for 2 hrs and see if it crashes, then you know whether it is the HD or not.

If it does crash then see if you have 2 sticks of memory, if so - remove one and re-test, then swap them over and re-test...if it stops crashing then you know it is the memory.

If it still crashes then your motherboard or cpu is probably fried, replace.

Hope you sort it out,

Nigel.

This a good choice?

http://slax.hosting4p.com/download.php

[95TBird]

Phill:

What size is your power supply, what brand is it & how old is it ??

This sounds like a power supply problem to me. See if you can borrow one larger than what you have now & see if your problem goes away.

The seemingly "random" reboots, then it working fine for awhile etc, would indicate to me that you either.

1. have a failing power supply

or

2. have a component that is randomly drawing too much current (shorting etc.) & causing the power supply to fail.

If it's #1 borrowng a power supply will tell you right away.

# 2 is a bit harder but just pull all the components (tuners , extra drives , etc) & then replace them 1 at a time til it starts failing again.

But I'm betting on #1

If it does end up being the power supply, get a good brand name one that is bigger than you need!! It will leave you room to add drives etc in the future & make sure that your machine runs cooler.

-Bruce

[PhillJones]

It's an 18 month old Seasonic S12-380.

So it's not too old and it's loads more power than I need, especially for such a well made PSU. It could be a bad one I suppose.

[PhillJones]

Originally Posted by Article22

It may or may not help but you can often get these kind of errors with memory, motherboard or cpu problems.

If a HD goes you generally know about it (grinding noises from HD, and nada ever again) which you haven't had...

I would take out any internal PC cards bar the graphics card (if you have one), set the bios to boot from cdrom first (here's hoping you have one) then boot the pc up with a Linux Live CD which has memtest on it, and run it for 2 hrs and see if it crashes, then you know whether it is the HD or not.

If it does crash then see if you have 2 sticks of memory, if so - remove one and re-test, then swap them over and re-test...if it stops crashing then you know it is the memory.

If it still crashes then your motherboard or cpu is probably fried, replace.

Hope you sort it out,

Nigel.

Okay, so I found the man page online for memtest and don't understand how to run it. I used to use linux but never really got the hang of it.

[PhillJones]

Originally Posted by PhillJones

This a good choice?

http://slax.hosting4p.com/download.php

That didn't work.

"-bash: memtest: command not found"

I guess slax live doesn't have memtest. Can anybody help me find a live .iso that has this feature?

[gilded07]

I've dabbled with both BartPE and Knoppix. I don't know if either has a memory test (Knoppix might) but google them - it's worth a try.

[PhillJones]

Originally Posted by gilded07

I've dabbled with both BartPE and Knoppix. I don't know if either has a memory test (Knoppix might) but google them - it's worth a try.

knoppix doesn't apparently. I have UBCD4win, which I used for the HDD swap out and restore. It's like BartPE on steroids. I think it has some kind of memory test on it but I assumed that Arioch was suggesting a Linux distro for some particular reason, like you have better control over your system under linux so can actually believe the results, or something.

I'll check UBCD4win rather than all this mucking about with linux. I reckon I'm too dim for linux anyway.

[doc]

I'd keep checking your hard drives, but if you have two 512MB memory sticks, take one out and see how it gets on.

If it still crashes, put the other stick in and remove the first.

Bad memory sticks will cause random BSOD's, and the error messages it gives you usually change every time.

You need to get to a known good point and then start changing things from there. Not easy when you don't have lots of spares lying around I know!

[Arioch5]

I've used the Seatools before. If you're drive has been over temperature it will tell you that when you run the test (non-destructive test). The reason I brought that up is I had 2 drives fail in the same computer that threw over temperature warnings. Neither one made an odd noise, but having them in the system (especially the system partition) would cause all sorts of random hangs and reboots. One of them wouldn't cause much problems b/c I wasn't using most of the disk, but once I tried to load it up with my recordings windows would lock up or reboot.

It's easy and cheap to test all the drives and if it shows no errors you've eliminated one source of error. As long as your hard drives aren't bad you're data is relatively safe so you can relax a little too

Unfortunately, if it's not a HD then it does become a lot harder. I'd check the power supply that sounded like a good suggestion to me. When I had a CompUSA PSU fail on me (only 2-years old) it did it gradually and caused the freezing and occasional reboot. The memory suggestions are of course good, if not a little harder to pin down. I typically test that last b/c I hate having to pull out all my hardware and add them back in one at a time.

P.S. I wasn't the on that suggest linux, simply b/c I'm a noob with linux :P

[PhillJones]

Originally Posted by doc

I'd keep checking your hard drives, but if you have two 512MB memory sticks, take one out and see how it gets on.

I don't have any memory sticks, at least not any permanently plugged in.

[Arioch5]

Originally Posted by PhillJones

I don't have any memory sticks, at least not any permanently plugged in.

In your signature it says "1Gb memory"

That's why he was suggestion if that's 2x 512 you can run at 1/2 ram to test if it will boot.

[PhillJones]

Originally Posted by Arioch5

I've used the Seatools before. If you're drive has been over temperature it will tell you that when you run the test (non-destructive test). The reason I brought that up is I had 2 drives fail in the same computer that threw over temperature warnings. Neither one made an odd noise, but having them in the system (especially the system partition) would cause all sorts of random hangs and reboots. One of them wouldn't cause much problems b/c I wasn't using most of the disk, but once I tried to load it up with my recordings windows would lock up or reboot.

It's easy and cheap to test all the drives and if it shows no errors you've eliminated one source of error. As long as your hard drives aren't bad you're data is relatively safe so you can relax a little too

Unfortunately, if it's not a HD then it does become a lot harder. I'd check the power supply that sounded like a good suggestion to me. When I had a CompUSA PSU fail on me (only 2-years old) it did it gradually and caused the freezing and occasional reboot. The memory suggestions are of course good, if not a little harder to pin down. I typically test that last b/c I hate having to pull out all my hardware and add them back in one at a time.

P.S. I wasn't the on that suggest linux, simply b/c I'm a noob with linux :P

Thanks for the clarifications on the HDD test. If it tells you whether the disk has ever been over temperature, I guess the test I just ran on the system disk rules out the possibility that the system HDD is overheating.

I'll check the second one tonight.

PSU is harder, I'll have to hunt one down. I might do the memory test first as I only have the one pci card, my capture card. The only other thing in it is the graphics card and HDDs. Th HDDs are easy to quickly disconnect. The PSU however is a knightmare as the power connector is between the huge graphics card heat sink and the huge PSU heatsink, which consists pretty much of a bunch of razor blades all lined up in a row.

[reboot_this]

Phill, I had very similar problems with my dedicated HTPC a long while back... my power supply started dying and the problems seemed to get more worse and occurred more often. Replaced it with a psu twice the size (for future-proofing) and haven't looked back.

Short story: I checked things like ram, video card, hdds, etc by swapping stuff out. I even completely reinstalled Windows and other software. Then one day I was at Fry's, and ran into a nice fellow that gave me a very detailed "class 101" lesson on power supplies... Needless to say he nailed it based off the issues I described I was having in the first place, it just would have been nice to run into him before I did all the swapping/testing/reinstalling of things.

[PhillJones]

I've put the old system disk in after testing it under seatools for windows as a second drive (well third actually but whoose counting) in my work rig. I've also tested the media drive using seagate's boot disk.

The system booted up again thankfully and I've been able to turn off autoload of sage at startup (a complication as sage won't start and raises all sorts of merry hell if a HDD isn't there.)

The situation as I type this is that the dirty bit on drive E: got set when the system crashed two nights ago so I'm doing a manual chkdsk to clear that, it's taking a long time do do stage 4 (verifying file data), which worries me, but maybe it's normal. After that's done I'm going to run seatools on the system disk again while it's in situ and then I guess I'll move on.

I think I'm goig to fire up the UBCD4win disk and run the memory test without taking anything out because knowing my luck, I'll take everything out, run the test it'll pass, then I'll reinsert things one at a time and re run each time and it'll pass everytime. I'll do an end run on this by testing whether it's going to fail at all to start with.

Now, I've disconnected and reconnected everything so if it passes memtest I'm going to try and run it again and see if the problems persist. If they do, I'll swap the PSU.

[PhillJones]

I ran memtest86, which is on the UBCD4win and got some results.

Now before I strip the PC down, I wanted to see if anybody could shed some light based on the rather odd result I got.

The standard test found nothing but when I changed the way that memtest got the memory locations from e820 - std to e820 -all, which includes the 'bios reserved memory' I got thousands of errors per test and memtest 'halted' while performing test 4

Test 4 [Moving inversions, random pattern]

Test 4 uses the same algorithm as test 1 but the data pattern is a random number and it's complement. This test is particularly effective in finding difficult to detect data sensitive errors. A total of 60 patterns are used. The random number sequence is different with each pass so multiple passes increase effectiveness.

I couldn't get a log file because the programme halted and exiting caused a reboot.

So, does the fact that I only get errors when tesing the 'bios reserved' memory actually mean someting in of itself.

I ask before I start taking components out because once upon a time I got too screw driver happy with a PC I was trying to fix and through a series of mistakes ended up killing the PC completely. Since then I've been very conservative about this sort of thing.

Cheers,
Phill

[unkyjoe]

P.S If you have any bad sectors on an IDE or SATA drive, it means it is failing.

I agree with checking the PS unit, the easiest way is to replace it with another unit, I always keep a spare on hand just in case.

Also when you get windows up and running, check your event logs and make sure you are not seeing any disk errors, I have had a drive pass the manufacturers test too many times and then had errors in windows to trust the man HD tests.

I would bet on a bad HD or PS.