Locate and fix hardware faults
Reporter
Once Checkbox has run all the required tests, the software displays a pie chart in the program window that represents the passed, failed, and skipped routines in absolute terms. A report, which is generated by clicking on Save Detailed Report at the bottom of the window, supplements this less than useful display. The report lists the completed test routines in neat groups and is stored in the Documents
folder in your home directory.
In the Comment section, the software provides some notes on the outcome of individual routines, so you can find out which test runs have caused problems. This enables detailed conclusions on the support for, and functional capability of, individual tested components (Figure 10).
Stress
The Stress program lives up to its name [15]. This small command-line tool tests the limits of hardware components by artificially generating load. The software focuses on the CPU, as well as the mass storage and RAM of a system. Stress is available from the repositories of virtually all major Linux distributions and can be installed in the package manager.
The man page or output from the stress -?
command informs you of the software's command syntax. The output also shows examples.
Stress is a very simple tool; it doesn't generate any reports and is quite reserved when displaying information in the terminal. It is always advisable to specify the -v
parameter and observe the load on the system in a second window via top
or htop
. For example, the command
stress -v --cpu 4 --io 4 --timeout 20s
generates a full load on the CPU and the I/O subsystem for 20 seconds (Figure 11).
Stress succinctly informs you in the terminal when the stress test is complete, with no further details on the generated workload (Figure 12). Because of its relatively simple routines and optionally configurable time out, Stress is especially suitable for intensive long-term CPU tests that simulate the stability of a computer system under full load.
Conclusions
The test and analysis programs presented here are very much heterogeneous in terms of their functionality and appearance. Other than Checkbox, you will almost always find the tools in the repositories of the major distributions, and the tools are all fairly easy to install. Thus, you can put them in place quickly on machines already showing signs of faulty behavior. (For additional analysis tools not tested in this article, see the "Untested" box.)
Untested
In addition to the tools discussed here, there are other tools that were not considered in this article, including the well-known benchmark Bonnie++ [16] and the equally widespread Phoronix Test Suite [17], among others. We did not consider Bonnie++ in our lab because it is biased toward mass storage benchmarking.
The Phoronix Test Suite provides many practice-oriented benchmarks for comparison purposes with other computer systems. These are based on application-specific performance measurement and are therefore less suitable for identifying problematic hardware components.
One of the benchmarks offered by Russia's UNIGINE LLC [18] is limited to graphics cards and is also application-specific. LuxMark [19], an OpenCL benchmark also primarily focuses on applications with a rating system and only targets graphics hardware.
The software tested for this article ranged from predictive analysis for mass storage to test routines that detect hardware defects. The components included benchmark tests that exposed a wide range of components to full load and thus, in part, helped to locate thermally induced or sporadically occurring problems. Regardless of the individual user scenarios, these analysis tools also ensured a more complete understanding of hardware interactions and functional capability. In this respect, you will not want them missing from any of your systems.
Infos
- inxi: http://smxi.org/docs/inxi.htm
- dmidecode: http://www.nongnu.org/dmidecode/
- SMART standard: https://en.wikipedia.org/wiki/Self-Monitoring,_Analysis_and_Reporting_Technology
- Smartmontools: https://www.smartmontools.org
- GSmartControl: http://gsmartcontrol.sourceforge.net
- DisKMonitor: https://github.com/papylhomme/diskmonitor
- Gnome Disk Utility: https://git.gnome.org/browse/gnome-disk-utility/
- "Linux and Laptops" by Erik Bärwaldt, Linux Magazine, February 2012, pg. 34
- e2fsprogs: http://e2fsprogs.sourceforge.net
- hdparm: https://sourceforge.net/projects/hdparm/
- MemTest86: http://www.memtest86.com
- Floppy disk version of MemTest86 v5.01: http://www.memtest.org/#downiso
- HardInfo: https://github.com/lpereira/hardinfo
- Checkbox: https://launchpad.net/checkbox-project
- Stress: https://people.seas.harvard.edu/~apw/stress/
- Bonnie++: http://www.coker.com.au/bonnie++/
- Phoronix Test Suite: http://phoronix-test-suite.com
- UNIGINE: http://unigine.com/en/company/profile
- LuxMark: http://www.luxrender.net/wiki/LuxMark
Buy this article as PDF
(incl. VAT)