ralf.mardorf at alice-dsl.net
Sun Nov 12 16:54:10 UTC 2017
On Sun, 12 Nov 2017 15:37:50 +0000, Ian Bruntlett wrote:
>Given personal experience of refurbishing computers for the Contact
>Computer Wombling/Refurbishing Project, I have found that when a
>computer I'm working on fails to run memtest86+ , that computer is not
>reliable enough for normal use. After a memtest86+ test has been
>passed, I wipe the hard drive using dban. If that fails, it is a
>strong indication that I'm going to have to change the hard drive.
I'm using Memtest myself as a point of reference to keep in mind, but
de facto RAM could be 100% ok after it failed and it could be borked
after it passed Memtest, let alone that one of the computers I tested
always failed when using Memtest from Ubuntu repositories and
always passed Memtest of the same version, when using a Memtest live
CD, both after shut down and disconnecting the PC from the mains.
There are tons of other possibilities that could cause issues, even
clearing CMOS _without_ replacing the battery could solve a lot of
Just running Memtest gains absolutely nothing. Running Memtest in
combination with other tests and actions is very helpful.
Smartctl is not perfect, but another link in the chain, even unplugging
SATA cables and connecting them again is important and so on and so
Just running one test, especially the free as in beer Memtest release
gains absolutely nothing, at least taking a look at /var/log,
journalctl and Co in combination with software testing hardware and
remounting, cleaning, etc. should be done. More experienced users could
use coolant spray etc. ...
Always keep in mind that measuring a faulty system by software running
on this system is fishy. A repair shop always tests equipment with
other equipment galvanically isolated from the mains.
Guerilla Open Access Manifesto:
More information about the Lubuntu-users