Ideas for diagnosing system hang

[FSin]

My system keeps hanging\freezing randomly (it seems). No input - mouse doesn't move, keyboard doesn't work - no ctrl alt backspace. I am going to try alt + sysRq and REISUB next time it happens to see if that works. found that on these forums. I am having to push the button so often I am worrying about my HDDs.

I think it is a hardware problem, I have a feeling. Also since I am now running Helena KDE 64bit, and previously I was running Elyssa KDE 32bit, and the same thing is happening still.

My feeling says RAM, but before it said GPU, and it wasn't that I don't think. So before I save up and splash out on some RAM for my very outdated system I was hoping to diagnose it somehow.

I have tried a few benchmarking tests, but I don't really know how to interpret the results. I also managed to find the /var/log folder (searching the forums), and had a look at those logs. I googled some dodgy looking lines but it all turned out to be normal enough. I would like to understand it - it seems useful information. But thats another days confusion.

So can anybody suggest any ways to diagnose the problem?

[remoulder]

before I save up and splash out on some RAM for my very outdated system I was hoping to diagnose it somehow.

Use the Memtest option on the livecd boot menu. If you get any errors it's faulty, simple as that.

[FSin]

Ok great I'll try that.

[DrHu]

My system keeps hanging\freezing randomly (it seems). No input - mouse doesn't move, keyboard doesn't work - no ctrl alt backspace.

Dtrace,. if available for your Linux is the best tool available..

• I'll just say it, takes a lot of work, and requires you know exactly what you are doing, although it won't harm the system to use it

http://blogs.sun.com/bmc/entry/dtrace_on_linux
http://www.cyberciti.biz/tips/dtrace-by-example.html

• dtrace a guide, plus other sysadmins tools to consider..

[FSin]

I tried memtest, and it turned up zero errors.
I just ran the automatic test, which did 8 different tests, once. It turned up no errors.

So it seems my memory may not be the problem.

I am not at my computer at the moment. I will be able to play with it again on Wednesday, so I will look at the DTrace thing then.
Thanks for the links DrHu!

[FSin]

Yeah. DTrace looks hard.
It is probably exactly what I need though.
I have been putting off using it - last thursday I spent a few hours reading about it, and even the simplest guides lose me at various points. I suppose if I keep reading about it then it will all start to make more sense, but I am not a software engineer or anything. I don't know the ins and outs of the linux kernel. Even if I did I suspect diving into it could be like swimming across an ocean in the night.

lol. poetic cheese.
i'm not even sure if its in my os yet. haven't checked. do that now i guess.

[det4100]

You didn't mention what kind of rig you are using. You did state it was an out-dated system. Some hardware problems are hard to diagnose. If you have more than one RAM chip on board, you can always pull one of them and see if it stops hanging up. Of course, swapping it out with known good ones would be better, but we are assuming that you don't have any on hand. Next, I would take a good look at all of the capacitors on the main board and on any sister boards and see if any of them look swollen maybe even cracked. The "can" style capacitors have weak spots in the metal casing to prevent them from exploding. That is where they usually crack open. Many motherboards where made with some bad capacitors that came out of China several years ago. I've seen quite a few of them. A bad capacitor can work just fine, for a while, until it starts to warm up. Heat is a killer. Gotta keep things cool.

Good luck with finding the cause.

det4100

[FSin]

Hi det4100, thanks for the input.
To answer your first question - its an asus\foxconn motherboard. gxk something i believe, and i think it would fit the criteria re coming from china, and several years ago. it has only the one 1gb stick of ram, ddr. its the only pc i have access to that isn't ddr2.
i checked the caps, and they all look swell, which is to say not swollen. roflmao, i kill me. since i don't have any spare ddr memory i swapped the stick to the other slot, don't expect it will make a difference but who knows.
As far as heat goes, the system is pretty cool really. I don't trust the old CPU cooler much now, but case temperature must be quite good -it has way too much airflow i did have trouble with the psu recently - which was that it was too loud. it is a very cheap psu :frantic:. i swapped out the fans for better ones and it is quieter now.

i probably won't do much on it till tomorrow or friday because i am like a zombie today. would definitely do more harm than good. although I may set up a check of the hard drives later since i can't mess that up too badly.

thanks for the suggestions

[FSin]

new type of crash today. 4 types total now.

1: original crash everything freezes, no input, screen does not return if power saving. ctrl+alt+backspace does nothing. Hasn't happened for a little while so I haven't tried alt+sysRq REISUB (only learnt of this recently)

2: browser crash, everything inside browser freezes, all windows, tabs, menus etc. can kill process and recover, sometimes if i right click and choose close from the task bar\panel icon it will work if i give it a while (say 10 minutes?) to deal with the request. Mint\KDE will also offer me the option to kill if i do this. Big CPU usage. Effects all browsers.

3. browser crash(?). all windows disappear in an instant. vanish. can reopen and recover session. This has also happened to other running programs, they just instantly close, hence the -(?). No effect on the system generally.

4: new crash (yay): using open office, twice in last 40 minutes, the whole system freezes up, time stops ticking (in PC), but I can still move the mouse around. Cursor icon remains as was in openoffice. Ctrl+alt+backspace does nothing. waiting does nothing. alt+sysRq+REISUB reboots.

badblocks tests say testable HDD partitions are good.

EDIT: more details about my system:
AMD sempron 3000(*?)
ASUS\Foxconn 760GXK***...?
1024mb RAM DDR x1
2x 80gb HDDs (seagate [master,boot] + western digital [slave,ntfs storage])
ATI X850XT AGP video card
Generic PCI sound card
Cheapo 450W PSU
LG SuperMulti Combo drive.

[JoeFootball]

My system keeps hanging\freezing randomly (it seems). No input - mouse doesn't move, keyboard doesn't work - no ctrl alt backspace.

For what it's worth, I had a similar sounding issue which I was able to resolve via this posting.

Joe

[FSin]

I can cause the crash at will now. not with my mind. that is not a great sentence really.
I mean I can recreate it whenever. The technique I am using to do so is to resize a picture using the mouse in an open office writer document. ie. dragging the corner of the picture = instant crash.

I can 'alt+sysrq+r' and then 'Ctrl+alt+del' (but nothing else, eg 'ctrl+alt+esc', 'ctrl+alt+backspace', 'ctrl+alt+f1', etc)

checking /var/log/messages, and syslog (if i recall correctly these were the more useful ones), I found logs of the sysrq calls, which is handy to isolate when the problem occurred. Immediately before these commands, usually there is a gap in logs of a minute or so. 30 seconds maybe. Which is probably the amount of time I waited while the PC hangs in hard freeze before I forced it to restart.

I wonder if there is anywhere I can check logs for that unlogged patch of time. ?

Or if there is anything else I can do. Reseat my CPU maybe... ? That fan could use replacement too.
I don't know where to begin with DTrace.

Maybe I should try a different kernel or something

[FSin]

Installed Mint 8 XFCE yesterday. its 32 bit I think, same kernel as far as I know.

I can still recreate this crash by resizing an image in an Open Office Document, I am not sure about the other crashes yet, since I don't know how to cause them. I don't think its related to Open Office.

[beefstu]

I was going to suggest maybe it was a bad install but if you've done a fresh install of XFCE then it can't be that! Could be an idea to try an updated kernal, they're a level 5 update in mint update, you need to make them visible in the preferences menu.

Certainly a bit of a mistry all of this! I suppose it may be worth keeping an eye on temeratures if your worried about he CPU. There's things like xsensors in the package manager and there's a temperture monitoring app that you can add to the panel in gnome but can't quite remember the name. No idea if it would work in XFCE too.

[det4100]

These symptoms you describe sure indicate a hardware problem. I've never seen a problem while resizing a photo from within Open Office. However I have seen problems, due to graphics, of resizing the Open Office window itself. I don't know how well ATI graphics are supported anymore as I tend to stay away from them when building Linux boxes. I never could get them to work right without researching for problems. You mention that you cause this crash by resizing an image from within Open Office. Just curious here, what type of image file is it? Is the aspect ratio locked to stay the same. I'm wondering if the graphics card is having trouble stretching and compressing the image as it is being resized. I wouldn't bother reseating the CPU unless you believe that that there may be a problem with heat transfer to the heat sink. I would maybe try reseating the graphics card, if indeed it is a card and not on board graphics. If it is on board, then maybe check to see if the heat sink is loose on the chip. Sure wish you had another stick of RAM to stick in there. I've swapped plenty of RAM that came up good on the test but failed in certain boards.

Good luck, wish I had an answer for you.
det4100

[beefstu]

To try to see if it is the graphics, could be an idea if you have onboard graphics too to try removing the dedicated card and just trying the onboard. May have to change a setting in the bios and I'm not too sure how Linux deals with changes like that, may need to change xorg I guess.

[mick55]

Hi FSin

Run these 2 commands in a terminal and copy/paste the results here.

This will tell us your hardware

Code: Select all

inxi -F

This will tell us what loads at boot time and show any errors

Code: Select all

dmesg