Hi all, I've been running my personal server for a year now. I've tried a bunch of distros.

Fedora Core1, then Suse 9.1, then came back to Debian Sarge.
But I'm starting to get very frustrated.

Why can't it be stable for once? I've had my setups checked by several people everytime, and I looked up problems & all, and my current installation is, should I say, spick & span. But, just like everytime I used a different distro (previous ones had some bugs that were really pissing me off, and Debian just works so well... I can't believe I still have things like this happening..)

So either the machine rebooted every rand() days, either it became unresponsive in the past. Generally, that meant blackscreen with no keyboard/mouse feedback. One time, with debian, when I hard reset, I saw messages saying that... it was ignoring a pci display device... uh yeah great, looked like it decided that my video card would be pci instead of agp )

Right now it's in coma. When I came home it was in an unresponsive state. Hard reset, thinking "oh well, it's going to sort up after a minute or two I guess". It did not.

The machine seems cut off from the world (I did check CPU usage and run an ifup eth0). Gaim won't connect (yeah, right "why the heck d'you need a gui in a server, and why do you need gaim?" Well I just tend to be really faster in guis, and Gaim does help me quickly check if the server is up), apache and webmin and all seem dead, and even firefox won't start properly =_=



And I wonder. What the hell did I do? I mean, hey I'm not doing anything out of the ordinary http & ftp servers, how can MY machines mess themselves up? I can't have them work for more than two/three months straight without something booming itself up. How do professional webhosts manage to have servers running for YEARS? I've never been able to exceed 43 days, I think. Always some trouble to fix. Well at least I don't get viruses right on the desktop like I did, once with windows 2000 ^__^

Sorry if I may have let some words slip out, I'm kinda getting annoyed by my lack of power in situations like this (else I guess I wouln't have to write all this ^^)

So, yeah... the server's down for now..

If you may think it's hardware failure, here are the specs:
¯¯¯¯¯
AMD athlon XP 2400+, I think I downclocked it to 2000mhz or somethin'
256mb DDR ram (can't remember what clock)
2x Quantum fireball (20gb each, at 7200rpm)
Ati radeon 7000, single screen
Power supply... uh can't remember, and since the case's off to allow some cooling...


I love linux and I'd really like to trash windows off forever. But sometimes, I really doubt that "to mess a linux box, you gotta work at it". It has been a tough year learning how to get all this working, but it has been really interesting (and yet frustrating also). I don't want to give up when I came to be so... almost good at it!


I will try to get some external help on my issue, if I can fix it out meanwhile I will post out my findings.

Hi kiddo,

Sorry to hear that you're having trouble with your workstation/server.

However, from what you've said, I think I would point my finger at a hardware problem.

As I'm sure you're probably aware, changing the clock speed on any CPU, up or down can cause timing problems. Why have you downclocked it anyway?

Does your PC really run hot enough to require this? Unless it's heavily loaded all the time, I would expect this to be a sign of trouble. Is your CPU fan working properly ? (see sig )

Linux is definitely a stable operating system, where I work we have a number of linux servers and have very little problems with stability with any of them. I can't recall any one of them crashing totally, despite enormous loads, or software problems. Which is pretty good when compared to the 200-odd windows servers that have to be split into bi weekly reboot schedules.

Still, one of the reasons that you'll see sysadmins able to keep machines up for years is the fact that this is exactly what server hardware is designed to do, all the components are selected not only for performance but for compatibility, many servers will have 2+ PSUs that can be swapped out without shutting the server down, the same with PCI cards, harddisks etc these days! So don't feel bad if you're having hardware problems (we have one machine which has an annual maintenance contract of more than my annual salary. It bloody better stay up!)

Anyway, one method you could try is removing all the nonessential components, and returning the clock speed to normal etc, installing a plain default installation of Debian, and just seeing if that will stay stable. Is there nothing in the logs that could indicate a problem either?

But stick at it!

Cheers,

Steve

Hi, thanks for your quick reply! It's good to finally have someone else's view on this, since this is a *really* petty server compared to things we usually see in a server room (note that my server room is actually.. my wardrobe !)

The downclocking: well I think I downclocked it to reduce heat (amd... do I need explanation? ^^) I don't have cooling equipment other than a simple fan, and a case fan I jacked into the wall to suck the hot air out... heheh

I was forced to get this AMD CPU and a new motherboard when I actually got my P3 motherboard badly flashed (yep... stupid it is.. anyhow, I could'nt get it hotswapped or get a replacement, P3 motherboards weren't available around anymore)

I will try getting back to automatic speed. Don't really think this will change anything, since I believe I had downclocked it for the same reasons. It's a shame that I don't have thinkgeek's laser-guided thermometer =)



About heat & load: Nay, the server really is not loaded. Below 1.
Having electronic parts heat is a little bit normal without powerful cooling I believe.
I remember, *BEFORE* I put my 120mm ball bearing fan to cool the room, I had excessive heat problems. Like, glue from the powersupply dripping on my radeon ..lmao (yeah it's scary, but no problems anymore.. and it seems it did not do further damages)

Since I installed the "air duct", the "server room" is quite cool now, I just checked at the ceiling level, with my super-technological tool (my hand), it's the same as the ambiant temp in my current room... Let's say between 20 and 25 celsius. So there's no overall heating in there.


About the logs: well I've tried finding something that could give me some clues, but I don't really know where to look. Where should I see the adequate logs?




The big problem in this is that it happens totally randomly. I could have nothing for two weeks after this post for example, then believe it's fixed.. then it could happen again.


//edit:
I just fired up the screen to check out the load (to be sure) and guess what.. No response.. Great. Note, there is no problem with my radeon or my monitor, since I tried pressing capslock on the keyboard. No LED responding. This definitely looks like faulty hardware! >_< I'm going to hardreset again ---- hmm no, poweroff should be wiser... and put the cpu clock to auto now.
Hope this info can help a bit.

--- oh well, while I'm in the bios.. I'm going to give out the health status:
CPU TEMP: 43 celsius (woah, that's almost cold o_o heheh)
MB TEMP: 31 celcius
cpu fan speed: 3358 rpm
vcore: 1.63 v
+ 3.30v: 3.24v
+ 5.00v: 4.84v
+12.00v: 12.9v

----edit
Okay, looks like it doesn't want automatic cpu clocking o_O didn't want to boot at all. Multiple tries. Anyways, put back to 100mhz, and it booted. Now I'm scared. Powering off.

Hey,

There's definitely something a bit weird going on hardware wise I reckon with your machine!

My machine at home lives in my "server room" in my lounge with all the other crap in there, and runs 24/7 no problem. It's an AMD too, I'd actually pick AMD over Intel for my PC. Just need that AMD64 Sure your PC might be hot, as long as it's not too hot!!

As for checking logs, most error messages will be written to /var/log/messages - I think this is the same on Debian anyway. It's also worth checking out the /var/log/syslog log file too.

Sounds to me like faulty hardware too. I would guess the motherboard, but I'm no hardware engineer.

I guess you can't borrow a motherboard that's definitely compatible with your CPU, Ram etc? Then strip the machine down and put back the bare minimum of components and see what happens, if it's stable then keep adding them and seeing what happens. Might be worth checking the CPU fan is working alright whilst you're at it! Hmmm, if I were you and I had a P3 CPU knocking around that had worked well, but just a duff motherboard, I'd be tempted to look for another MB on eBay or something like that, and give it a try too, and see what results you get.

I'm not sure I'm being particularly helpful actually, but good luck!

Cheers,

Steve

Well the CPU and board were spanking new, bought together, they'd better be compatible ! x_x (the board is an Asrock K7VT2). I think I've found out the problem, but I can't be sure, since I have no way to test it other than send the PC for a check-up at my fav shop those guys there sure know me by the time...

I decided to boot the server just to look at the logs but... Looks like it didn't want to. BEEEEP BEEEEEEEP BEEEEP (hmmmm...) I took out the 256mb PC133 ram, and there's something very fishy. I scanned it at 600dpi so you can have a look, too (and possibly laugh a little ^^)

http://www.nanokron.info/hophophop600dpi.jpg
File size: 273kb (this is my windows workstation as an auxiliary server... guess I'll be hacked in a matter of days)



Notice how it looks like it has been on a campfire (near the 2nd cavity)! Anyhow, I'll have to send the computer to check, and pray they find a way to sort this out.


As for the cpu fan... Well, it runs, how should it be supposed to run "correctly" (well.. in fact if it didn't run properly, the cpu heat would be higher than 45 :P)

Thanks for your input!

Where do you see damage on the ram cause as far as i can see it looks normal ?? I would probably reset your bios to defaults and dont under clock it anymore. Another thing is that the beeping your getting is a bios error code so theres something up with it hardware wize. 3beeps indicates <<SOMETIMES>> video card failure/post problem/keyboard error.
take your vid card out and see if it boots up. <-- I understand this sounds moronic but if you see disk activity like normal then its probably booting normally. Anway if you give me your bios inoformation and post exaclty the sequnce of the beeps i can give you a better idea of what it is. Good practice is to remove all hardware execpt the essensial stuff video/keyboard/ram/processor then clear the bios with the bios jumpter and attempt boot. If it boots great if not try changeing hardware out/ ex. diffrent ram/video card/keyboard <-- if possible of course. I wouldent personally take it into anyone cause this is all there gonna do anyway and charge you money to do basicly nothing. You seam like you know what your doing about as much as any computer repair person is gonna know. HAPPY HUNTING

Uh... whoops I thought I had posted those details:

- yes, I had already tried without the video card
- well I thought that having a dark-brown area on my RAM was kinda scary.
- don't have acces to the bios, since it won't boot up at all
- no problems with keyboards/mice either

I forgot to point out the particularly funny manual of the K7VT2 motherboard: it does *not* have a troubleshooting section! It doesn't even mention error codes! For now, the PC is being checked. May be the powersupply too, but I don't really think so. hmm, while I'm here, I'll just call them and see if they found out.

--- okay
They said it now boots, SEEMS like there was something blocking the AGP port's contacts. I'll go and check it myself later. If THAT was the problem, I'm going to get an automatic AGP-vacuum !

I don't really know how could dirt block up the AGP slot, since I never removed the video card.. I guess that is why the PC wouldn't boot even without the card.


I will edit this message to confirm everything. Thanks!

You reset the bios with the jumper not in the utility. I figured it was a agp beep code crap. I have this same issue on my mobo cause the little clip that holds the card in place is missing :P

Kiddo, why don't just do memtest, it will tell you whether the memory works. Beats paying guys in shop to do the same thing

Check out: http://www.memtest86.com/

Also, reset your BIOS to normal settings. If you have to downclock your machine because it's too hot, than you are in trouble and probably shouldn't be using it. And try no to switch between distros too much. Linux is linux, and stick with one untill you learn how to configure it. (just my 2 cents).

UPS, I just read it wan't boot at all... Try pullin memory out of another machine and giving it a shot Also, clean all the connections. Buy a can of compressed air and make sure there is no dust in connections.

Yeah, thanks but, how could I run memtest if I could'nt get to bring up anything on screen or boot ? Normally, I'd have done it myself :P (hm and I usually don't pay anything for check-ups at the shop )

I just hope the problem is as simple as it sounds however, makes me mad to have started up a big topic on it ... oh but that brings me back to my starter question: even if it can boot, the server will still be unstable (by the way, I'll have to test if it can behave correctly when booted), meaning that after a few days it might just freeze again.

The little "can't boot at all" issue was juste a little thingy ATOP my main problem. So I think I'm back to my first post.. except that I will be able to boot and check the kernel logs.


P.s.: I often read that computer cases are designed to have better air flow when they are normally closed. I would like to hear your views on this, since in my case I removed the two sidepanels to have air circulate freely between the components and the ambiant air (which is sucked out of the room anyways)

Well, faulty memory can cause freezing. In fact, if it freezes, it probably is hardware issue. So, once you get it up and running : reset bios settings, buy a better cooler, clean all the connections, make sure your memory is all right.

Better cooler, for the CPU I presume? (40 degrees are not *that* much..) well I could get an Aero Lite 7G (which is shaped like a Boeing plane reactor !) but... not sure that is the problem. I shall do a memtest, too.

I'm still wondering.. any ideas why the monitor gets black when entering a frozen state, anyways?

Hi again,

Well, if there is a problem with the AGP port, maybe this is the cause? If your card isn't working properly, it probably isn't sending a signal to the monitor....and most go to standby unless they receive a signal. Or it's 13 years old like mine!

Cheers,

Steve

Okay, the server's up. I ran a memtest, no problems. There really seems to be some problems with the AGP thingy, but it's a matter of having it jacked-in the right way I believe. Let's see if it can stay for a normal uptime...

Okay. Back to square one. It is fu%?$% frozen again (less than a few hours uptime! wow..)

That means:
no video signal
no keyboard feedback
no ping
no services responding

silent machine consuming my electric power for nothing >_<
I'll reboot and add the logs here
---

edit: "A new challenger has appeared!"
I'm going to google it, but what the hell? Basically it looks like my eth0 couldn't be activated (it's not using dhcp)...
This prevents me from being able to access anything, can't do anything, control-C or Q or escape won't do anything, so I am jammed with this in the terminal >_< how come things like these can just popup for no apparent reason? I kinda feel doomed sometimes, unless there's something I didn't do.


edit 2004 11 14:
Yesterday night, I managed to get out the kern.log, messages, syslog, and dmesg. However, today, the mahine totally refuses to power up. F***