How to track a crash?

[Zweitaktmotor]

Hi all,

I hope I can find some useful advice in this forum. Here is my problem:

I have an Asus eeePC 900 with 1Gb RAM. It originally came with 4GB storage internally and an 8GB SSD (i.e. solid state, not a hard disk).

When I decided to upgrade from Linix Mint 9 to LMDE, I also though it a good idea to upgrade the SSD. That way I could keep my old Mint 9 ready in case anything went wrong. So I ordered a 32 GB SSD stated to be compatible with my eeePC 900

All went well and I am quite happy with LMDE, apart from the fact that I experience irregular crashes, about once per day.

This is what happens:

I have my system monitor in the task bar, showing CPU, memory and network traffic. Just before the crash happens, the CPU graph gets very busy. It is filled up 100% but only in dark blue (IOWait). The memory and network graphs remain normal. Then the graphs start to move choppy until they stop. I can still move the mouse pointer but it will freeze on the next click. My HD light remains constantly on. The only way to get out of this state is by doing a hard reset.

There is no particular indication when this might happen. I may be writing something, browse the Web or even leave the machine completely unattended. Normally, I have Firefox and Claws Mail open.

Obviously, it could have something to do with the new SSD, but between the crashes it seems to be absolutely fine.

On a physical hard disk with DOS or Windows, I would say it could be a bad sector and I would use something lilke Scandisk to find and mark it. But this is a SSD and I am using Linux, so I am bit lost.

First of all, can you point me to the correct log file please, so that I can find a hint what is actually happening when the CPU gets busy. There are so many entries in Log File Viewer and most of them are rather cryptic.


Many thanks!

[Gerd50]

Probably var/log/syslog could give you a hint what might be wrong. You can post the output here, if you need help with reading it.

[Zweitaktmotor]

Not really conclusive.

Last night, I went to bed leaving the macine running at about 1.30, this morning it was crashed again.

The last entry (before my reboot) in the syslog says:

Nov 25 01:17:01 asus /USR/SBIN/CRON[3388]: (root) CMD ([3388] cd / && run-parts --report /etc/cron.hourly)

I noticed several instances of this entry, always 17 minutes past the hour. What is it?

Many thanks.

[Gerd50]

It says there is a cronjob active each hour - http://en.wikipedia.org/wiki/Cron

What is the entry of your /etc/cron.hourly?

[Zweitaktmotor]

There is only one file called

.placeholder

# DO NOT EDIT OR REMOVE
# This file is a simple placeholder to keep dpkg from removing this directory

So it looks like cron is called up without anything to do? Can I switch it off?

[Gerd50]

Sorry, i know nothing about cron and whether this placeholder could be responsible for the crashes of your system. It's possible, if
this cron is defined to work as root. Maybe

Code: Select all

sudo crontab -e

can give you an answer whether it is like that. Or simply watch your system monitor what the CPU does if the cron is active.

[Zweitaktmotor]

Something else just occured to me (and if that's the case I am really sorry to have wasted your time):

It could be that the new component is malfunctioning simply because of overheating. I know the netbook was always a bit "warm", but how warm is "hot"?

I found a neat little utility called Jupiter that can display the machine's temperature. After installation, it showed 68 C.

Now I flipped the netbook upside down (can do that because I am using external keyboard and monitor) and pointed a desktop fan at it. The temperature went down to 53 C.

Let's see if it crashes again. If not, I need to get one of those cooler pads for the netbook to sit on.

In this case, I am very sorry, the problem would have nothing to do with LMDE.

Let's keep our fingers crossed.

[Gerd50]

No problem, you didn't waste my time. I found it interesting reading about cron and i want to know more about it now.

Yes, how warm is hot and what is too hot My last graphics card died when it reached 115 C. I could roast eggs on it
Well, i press thumbs and big toes for you that your netbook is only a bit too warm and there is nothing misterious in the
file system

[Zweitaktmotor]

No luck, the crash happened again, despite extensive cooling.

BTW, the cron.hourly also happens on my wife's machine and this one does not crash, so it might not be the cause.

Could there be a problem with the swap partition? What if I try to remove and re-instate it?

[Gerd50]

Too bad...reinstall the swap partition can't be wrong and what about your ram? Is it possibly flooded whilst a crash?
And are there no other hints in the syslog than the hourly cron?

[Zweitaktmotor]

I went through the pain and re-installed LMDE with all the updates in a fresh partition. It worked well for about 48 hours and then the crash happened again in the middle of me writing something.

I also changed and upgraded the memory to 2GB, so this is not the problem.

I gave the SSD hardware drive a check-over but could not find anything wrong. The manufacturer's customer service is not familiar with Linux: "Have you tried to install Window? Does it happen there as well?" No, I have not. Without any license for commercial software, it will be quite useless to me.

The bottom line is: Either LMDE does not like my hardware or one of the updates and additional software is rotten.

But I am really at my wits end. I just want to use my pc, not having to spend hours trying to find an elusive fault.

[Gerd50]

What about to remove the SSD for some days? If there are no crashes then you would know there is a conflict with it.

[Zweitaktmotor]

The SSD is my hard disk. If I remove it, I can only run live CDs (with all their drawbacks). There is no space or provision for a traditional HD.

I have exactly the same LMDE setup on my wife's machine, but this is a traditional desktop with a spinning HD. The crashes do not happen there. So I think it is some hardware incompatibility and/or failure with my netbook.

The SSD cost me the equivalent of about $100, but in order to make a warranty claim, I have to prove it is faulty and I was not able to do that so far. The manufacturer seems to think it is rather my fancy Linux OS that is the problem.

The current log files do not give me any indication what happens immediately before the crash. Are there any additional diagnostic tools I can install?

[Gerd50]

Sorry, i thought your ssd is a second harddrive. You can try smartmontools from the repos. But i'm not sure whether it will work
with your ssd. Possibly you can find informations on the web.

[Zweitaktmotor]

Dear Gerd50,

I just wanted to say Thank You for your patience and help.

It turned out to be a hardware problem. The manufacturer was not very helpful, but I have found an alternative brand and returned the faulty unit for a refund.

LMDE works perfectly now.

All the best, have a nice Xmas and "einen guten Rutsch".

[Gerd50]

Thanks for the news how it ended with your ssh. And the same to you with guten Rutsch und so