There's mainly 3 reasons on which the computer can fail:
- CPU: the brain of the computer, should compute correctly and enough cooled
- RAM Memory: sometimes has badblocks, very very bad for your data and the correct working
- Hard disk: sometimes has badblocks, can damage your data and turn your disk unusable
CPU
From the live or the installed system, run the command:
cd /tmp ; mprime-cpu -t
Leave it running for 20 minutes, if the process has stopped / exited is because it failed, otherwise you can press Ctrl + C to stop running it
Technically speaking, it runs a lot of processor calculations, stressing your processor and becoming at his hottest level, if the temperature is not enough well cooled, one of these calculations will return a wrong result and the process will stop telling you that failed
Do not worry about burning your computer, since the process stops in case of fail it will not continue being hot, otherwise even in full stress the computer is running perfectly fine
Solution
To have failed this test is really bad, you can experience probably random shutdowns in your computer but you can also have damaged data and crashes in applications, the best solution is to open the computer and remove the dust, call to a professional if you don't know how to do that
RAM Memory
Sometimes the RAM can have badblocks, to test it, in your Live or Installed mode run the "memory diagnostic tool" option that you can found in the first (grub) menu that you have when booting the computer. Leave it running during an entire night and see if there's any error
Solution
Remove the RAM slots one by one to found the one that has the errors (passing again the test), then replace it for a correct one.
Note: the RAM slots has different speeds and characteristics, is best to use the same ones for optimal speeds
Disk
Specially mechanical disks, it is very common that they will have badblocks, this will lead to data corruption and a probably soon disk unusable, remember to always have backups
The best way to test it is by running:
sudo palimpsest
And search for the "reallocated sectors" part in disk health, note that the tool can say that the disk is healthy and having badblocks (reallocated sectors), if you see more than a few reallocated ones, your disk must be replaced
To run a full test you can use:
sudo badblocks -vs /dev/sda | tee badblocks.txt
(sda is in this example your first hard disk, change it if you use a different one)
The test will take maybe an entire day to finish, if the file badblocks.txt is empty after it finishes you are fine, otherwise you have badblocks
Note: reiserfs v3 allows to format partitions skipping the bad block sectors, you can give this file to the tool to format it without using these sectors, make sure to use the same blocksize for that
Other disk causes
Maybe the disk has some mechanical issues too, you can run:
sudo iotop -o
To watch the disk usage tools
Also put your ear in the disk to hear what it says (yes, im serious), he can complain about some pain or you can listen some strange anormal noises
Solution
Replace your disk, we strongly strongly recommend to use a SSD hard disk, it is a hundred times faster and it doesn't give problems in a minimum of 10 years, we recommend intel or samsung trademarks for them, not other ones!
Graphic Card
Even if you don't believe this easily, a broken graphic card (or maybe its simply driver) can make a computer entirely block / freeze / crash, so if you experience these issues with your computer and none of the previous solutions worked, try to change your graphic card to another one, it will be easy to know if it was or not the cause
Tips
A good way also to know about hardware problem is to read the kernel message, specifically run:
dmesg
It will show you some kernel messages specially related to hardware, you may found useful info on it
Specially useful to run when you see that a problem just happened
You are done!
These 3 tests guarantee the main good health of your computer
Unfortunately there can be other secondary causes, less common, but for the sake of the good information let me include them:
Internet disconnections
It can be to a bad driver of your wifi (or network card, but that's not much probable), try to use a different one
Desktop / Graphical / Screen issues
The cause is the same: your graphic card, it can be due to a bad driver or even in a broken card (like its memory badblocks), but in any of the cases just change it to a different one. We recommend to use:
- Intel graphic cards: normally they are the internal cards in the computers, they are not very powerful (for gaming) but they are very stable and works very well
- Nvidia: even if needs privative drivers, they works good most of the times
- ATI (AMD/Radeon): avoid them, they are commonly very problematic with different issues
Suggested computers
Do you want to buy a new computer? that's easy, we recommend a single trademark for them:
- thinkpad for laptops
- thinkcentre for tower computers
No joke! these are simply good quality and real computers, everything else in the market are more like toys and not a real computer. And you can buy them as second hand and still very good machines.
I have recently buyed in Mexico a Thinkpad T460s for only 400$ and is the best laptop that I ever had, in every sense! it's just amazing
UPDATE:
Since 3.0.6, the health of the disk is automatically reported to the user: Disk health reporting features
