Blog: S.M.A.R.T. Tests

There are often questions about S.M.A.R.T. (Self monitoring & Reporting Technology) and what the reporting means. Some people have the misconception that it will always warn before a hard drive fails, therefore there is no need to run any sort of diagnostics on hard disk drives anymore. This is not the case as explained below.

S.M.A.R.T. is not always correctly implemented by manufacturers as there is very little that is part of the ATA specification, but much more that is proprietary and understood only by the manufacturer of the drive. Additionally, the correct manufacturers’ tool must be used with that manufacturers’ drive; e.g. using a different manufacturers’ tool on a different manufacturers’ drive will produce unreliable results.

Within the specifications there are a few common points that can be used. There is a “Global Status” that will indicate a failure if a threshold is exceeded: this will not tell you which threshold so it can only be interpreted as “this drive is failed” or “this drive will fail imminently”. This information comes from logic that the manufacturer has implemented and relies on their defined thresholds and algorithms so it can only be taken as a warning.

S.M.A.R.T. also allows for the drive to be asked to perform a “Self Test”, a “Short Test” or “Extended Test”: these again use logic provided within the drive and produce a PASS or FAIL according to the manufacturers’ criteria.

There are many other thresholds within the proprietary information that can be read, however the interpretation is difficult without the manufacturers’ detailed information.

The implementation of the S.M.A.R.T. tests within Pc-Check checks the global status and carries out the “Self Tests” (if supported). Additionally, it also reads as many of the registers as are available and displays the values and the thresholds. As values always move down to the threshold, Pc-Check can determine if a threshold is exceeded for these registers. Pc-Check reads the S.M.A.R.T. logs and as every entry is recording an error, it will show the test as a failure if the log contains more than one entry per Gigabyte. To bring this in to context, Eurosoft has had cases where the S.M.A.R.T. log contained 1500 entries on an 80GB drive, but the manufacturers’ diagnostics still returned a PASS for the hard disk drive.

S.M.A.R.T. technology is a read only mechanism: it is not possible for the diagnostics to have any direct affect on the S.M.A.R.T. data and attributes held by the drive. However, the act of testing a drive will have the side effect that the drive updates S.M.A.R.T. data based on the fault conditions that it detects while it satisfies the demands of testing.

There is selective logic in the interactive mode of Pc-Check that aims to guide you. For example, if you run the S.M.A.R.T. immediate test and it shows a small number of errors: while it may not exceed the 1 error per GB rule and records a PASS result, it will then advise that you should run a S.M.A.R.T. extended test to confirm the drive does not have first indications of a serious problem that might then be captured by the more extensive testing.

All of the tests described above are found within Pc-Check, in the Windows-based product, Pc-Check Windows, some of the same tests are repeated and additional tests have been implemented to further test the device when it is not directly accessible as an operating system driver is present. Both scenarios are necessary as one has to be certain that it is the physical device that is causing the issues and not the use of an incorrect driver.

For further details about hard drive testing between Pc-Check Windows and Pc-Check, please refer to the attached document in the appendix in particular to the section entitled Hard Drive Diagnostics: Testing Methodology.

Conclusion

While manufacturer-specific diagnostics obviously have proprietary knowledge of the internal workings of their individual products, on the surface, this would tend to make the argument that because these tests are written by the manufacturer they must be better; these can also be written to accentuate particular features of their individual hard drive, avoiding any “grey” areas that may exist. They also are written for particular models of hard drives making it essential that the correct tool for the model of drive to be tested is used as these will have individualised output log formats, which means that hard drive comparison between manufacturer and models cannot be accurately determined.

In a support environment, it is a requirement to be able to compare testing results between different drives and their manufacturers with common standards and output format so that the performance between individual manufacturers’ products can be evaluated. It is important that in the case of returned suspect hard drives (unlike those tested in manufacture), the 100% default coverage should be used and time allowed for that. Testing with S.M.A.R.T. both before and after testing is another good approach because it may save time on some drives, picking up obvious failed drives immediately, while S.M.A.R.T. (immediate) testing again at the end will show if thresholds have been exceeded during testing (for example, used up spare sector pool or too many recalibrations). Additionally, hardware diagnostics treat hard drives at the sector level and so logical software constructs held within their data, such as multiple boot facilities, partitioning, file system and operating system choices have no direct influence on testing or results.

Eurosoft diagnostics are written to test the hard drives according to internationally agreed standards, allowing comparison between all products. Pc-Check base memory requirements have been reduced (i.e. increased available space for any 3rd party boot tools used so not to produce ‘Error’ test results); also increased support for SATA systems AHCI SATA (extends all SATA support to CD and DVD drives).