Hard Disk Monitoring

Hi,
first of all, I hope it's okay to put several bug reports in just one posting. If not please notify me for next time

1)

HD Sentinel uses the current value for any field (not only power on hours count) for many data points from the past. Bug or broken log file? Or a small reminder to buy a license? Data seems to be saved correctly, as they miraculously show up weeks/months later (once older than the bugged area)

2)

When switching to normalized SMART values, the explanation for fields with no (or no negative) changes is not displayed properly. If HD Sentinel is able to calculate a date of failure,it is shown correctly.

3)
(full screen example...btw, that drive really anticipates a trip back to WD!)
-> http://www.abload.de/image.php?img=hdsentinel3rtrt.png (BB-Code URL is prohibited?!) / http://www.abload.de/image.php?img=hdsentinel5jow8.png

On the overview tab, space is not used in an efficient way. While the status field is too small (especially with more than one warning message) and shows scroll bars, the time view underneath it is huge for no apparent reason.
Also, when scrolling backwards in time, the scale hops around. When there are two or more different values, they are shown as min and max, while when there is only one value it is shown in the middle (even if it is 100%!) As soon as values larger than the current max come in, the scale resizes. After scrolling back, it resizes again and shows possible smaller values now as high as the previous data. Confusing.
One more thing: As you can see in the screenshot, sometimes there is no explanation at the y axis. Sometimes there is...

/edit/ gee, missed one!
4)

Data entries for short ("Kurz") and long ("erweiterter") SMART offline tests are interchanged. Short test took me 15 minutes via the WDC tool until it was aborted. Long test took a few hours and displayed some error after completing the read test. Guess it was not recognized as a valid result and therefore SMART data was not updated by the HDD.


Bye and happy fixing
Bzzz

Thanks for the bug reports and your thoughts, of course no problem for using a single post !

1)
I can make sure it is not a kind of limitation of the free version.
It seems the database file for the statistics is broken/corrupted.
If possible, can you please
- use Report menu Send test report to developer option
- use Configuration -> Update -> Backup configuration and send the file to info@hdsentinel.com ? (but it is only possible in a registered version)
This way I could examine your database and verify the situation.

2)
I'm not sure if I understand correcty.
If the fields has no (or no negative) change, than it means that the attribute VALUE field is stable.
If the VALUE does not change (in negative direction) then it will never cross the lower line (the THRESHOLD line) so it cannot be used to predict any date of failure for that attribute. However, when you examine that dates, you may see that some attribute may only reach the threshold after 20-50 years (or even more, depending on the drive, attribute and situation). This is not the problem of the software but the problem of this failure prediction and disk diagnostic method, this is why it is NOT preferred. This is described with more details on www.hdsentinel.com/smart

3)
I agree that the drive may need to be replaced as it is in really bad condition.

> BB-Code URL is prohibited?!

Yes of course. Before that every day the forum was filled with 1000's of fake posts with only links to hmm.. offtopic sites (xxx/gambling/etc).

> On the overview tab, space is not used in an efficient way. While the status field is too small
> (especially with more than one warning message) and shows scroll bars,
> the time view underneath it is huge for no apparent reason.

Excuse me if you do not like this. On most drives (with fewer problems) the text
description is much short so displaying a big (but empty) green area would not be better.

> Also, when scrolling backwards in time, the scale hops around.
> When there are two or more different values, they are shown as min and max,
> while when there is only one value it is shown in the middle (even if it is 100%!)

Yes. The graph is showing the possible min/max values for the selected time frame
(the past days, if the software was active and logged the health for the particular days).

This may be different from time-to-time (if the health of the disk is changing)
and more preferred than displaying a fixed scale where slighest change of health
(eg. change in 1-2%) may remain unnoticeable.

> As soon as values larger than the current max come in, the scale resizes.
> After scrolling back, it resizes again and shows possible smaller values
> now as high as the previous data. Confusing.

Sorry if you do not really this method. This may be changed but the main idea
is displaying any possible CHANGE of the health with time to make sure the
user notice it.


> One more thing: As you can see in the screenshot, sometimes there is no
> explanation at the y axis. Sometimes there is...

Excuse me, I can't really see what do you mean. The Y axis always displays
the health as you can see on the tab just above the graph (Zustand %)


4)

> Data entries for short ("Kurz") and long ("erweiterter") SMART offline tests are interchanged.

Of course they are not

> Short test took me 15 minutes via the WDC tool until it was aborted.
> Long test took a few hours and displayed some error after completing the read test.
> Guess it was not recognized as a valid result and therefore SMART data was not updated by the HDD.

The estimated value (which were provided by the manufacturer) for these tests applies for a perfect
drive or a drive with has much fewer problems AND applies only if the drive is idle during the test.
Any other disk operations (caused by the OS or by any other software) may increase the test time.
So if the disk is not idle (especially if it is a system disk) and/or has problems, it is possible that the
tests may run for much longer (even 10-20 times longer) than the estimated value.

Thanks for your thoughts and response, I'll keep them in mind

hdsentinel wrote:Thanks for the bug reports and your thoughts, of course no problem for using a single post !

1)
If possible, can you please
- use Report menu Send test report to developer option

Done

hdsentinel wrote:2)
I'm not sure if I understand correcty.
If the fields has no (or no negative) change, than it means that the attribute VALUE field is stable.
If the VALUE does not change (in negative direction) then it will never cross the lower line (the THRESHOLD line) so it cannot be used to predict any date of failure for that attribute.

Exactly. And the message for this case is not displayed in full length, but chopped. Not even "Voraussichtlich" is shown (and that word doesn't make sense with no additional explanation *g*)

hdsentinel wrote:3)
I agree that the drive may need to be replaced as it is in really bad condition.

> Also, when scrolling backwards in time, the scale hops around.
> When there are two or more different values, they are shown as min and max,
> while when there is only one value it is shown in the middle (even if it is 100%!)

Yes. The graph is showing the possible min/max values for the selected time frame
(the past days, if the software was active and logged the health for the particular days).

This may be different from time-to-time (if the health of the disk is changing)
and more preferred than displaying a fixed scale where slighest change of health
(eg. change in 1-2%) may remain unnoticeable.

That's correct. But you do see the point of having a comparable (fixed) axis scale when scrolling, right?
Also, the values for single data points are shown above the graph.

What about adding a coloured filling underneath it, depending on the calculated healthiness? Any change in value could then also be perceived as change in colour. (I know, that feature would require some work )

hdsentinel wrote:3)
> One more thing: As you can see in the screenshot, sometimes there is no
> explanation at the y axis. Sometimes there is...

Excuse me, I can't really see what do you mean. The Y axis always displays
the health as you can see on the tab just above the graph (Zustand %)

Sorry, the HDD used for making the previous screenshots is no longer here, so I cannot access data from HD Sentinel for this specific drive. However, my current "backup" one that is also more dead than alive also shows this. But is hasn't that much data points, so the only screen I can show you displays no change in values. However, I can assure you the right one was also was the case for the old drive with a rapidly decreasing 100% -> ~20% graph.

-> left one: no description on y-axis, right one: huh, it's there!

hdsentinel wrote:3)4)The estimated value (which were provided by the manufacturer) for these tests applies for a perfect
drive or a drive with has much fewer problems AND applies only if the drive is idle during the test.
Any other disk operations (caused by the OS or by any other software) may increase the test time.
So if the disk is not idle (especially if it is a system disk) and/or has problems, it is possible that the
tests may run for much longer (even 10-20 times longer) than the estimated value.

Estimated times are displayed correctly, as you can see. No error in this.
But:

hdsentinel wrote:3)4)

> Data entries for short ("Kurz") and long ("erweiterter") SMART offline tests are interchanged.

Of course they are not

> Short test took me 15 minutes via the WDC tool until it was aborted.
> Long test took a few hours and displayed some error after completing the read test.
> Guess it was not recognized as a valid result and therefore SMART data was not updated by the HDD.

As you can also see in the screenshot above, 15 minutes are the displayed duration time for the long test. And that was simply not the case because it was the short test that took over seven times as long as predicted. Look at the date/time on top of that, it says 05.10.2010 (dd/mm/yyyy). The value for long test wasn't updated since 23.09...only WD knows why.
I guess HD Sentinel just reads both values (or maybe three of them, as the conveyance test doesn't show time information?), sorts them and then assigns the smallest value to the short test, and the largest one to the long test

Thanks for the report!

Bzzz wrote:Exactly. And the message for this case is not displayed in full length, but chopped. Not even "Voraussichtlich" is shown (and that word doesn't make sense with no additional explanation *g*)

Thanks, I'm checking and fixing that. Sometimes non-english phrases may be longer than english ones and may not appear as should. Excuse for that.

Bzzz wrote:That's correct. But you do see the point of having a comparable (fixed) axis scale when scrolling, right?

This is what I mean that on a fixed axis scale a minor change in health may remain un-noticeable.

Bzzz wrote:Also, the values for single data points are shown above the graph.

Yes, it may happen if the data point is placed on the top of the graph: the value is always placed above that point.
I do not think it is a problem.

Bzzz wrote:What about adding a coloured filling underneath it, depending on the calculated healthiness? Any change in value could then also be perceived as change in colour. (I know, that feature would require some work )

This sounds interesting I'm not sure if it can help in any ways to understand the change or help users to see clearly what happens but maybe... I'll think about it!

Bzzz wrote:-> left one: no description on y-axis, right one: huh, it's there!

I can't really feel it is a problem.
There are points with their values displayed on the graph. As you noticed, the Y-axis displays descriptions for some points (eg. 100, 80, 60, 40, 20, 0) if applicable. If the graph shows 59 and 54, there may be no Y-axis description for those values.
You may think that the graph is automatically zoomed to display the currently selected values.

Bzzz wrote:As you can also see in the screenshot above, 15 minutes are the displayed duration time for the long test. And that was simply not the case because it was the short test that took over seven times as long as predicted. Look at the date/time on top of that, it says 05.10.2010 (dd/mm/yyyy). The value for long test wasn't updated since 23.09...only WD knows why.

I just noticed that you mentioned that you started the test with the WD tool, NOT with Hard Disk Sentinel.
So you should not be surprised if the software was confused, if you did not use it to start the test?
The disk reports for the software that the "test is running" or "test is completed with/without error", but not WHICH test was started.
If you would start the test by Hard Disk Sentinel, then the software (of course) knows WHICH test you were starting. But if you use an other tool, the software can only check the test duration and based on that Hard Disk Sentinel saves the test result to the short or extended field.
So it is not a bug but a feature to be bullet-proof against user actions....

Again, Please use Hard Disk Sentinel for starting the tests because it provides more information about the results and logs the results so it can be used for future reference.

Bzzz wrote:I guess HD Sentinel just reads both values (or maybe three of them, as the conveyance test doesn't show time information?), sorts them and then assigns the smallest value to the short test, and the largest one to the long test

Of course NOT! Excuse me but you are wrong.
Hard Disk Sentinel reads and displays these values properly without applying such "tricks" on the numbers.

Hi

hdsentinel wrote:

Bzzz wrote:-> left one: no description on y-axis, right one: huh, it's there!

I can't really feel it is a problem.
There are points with their values displayed on the graph. As you noticed, the Y-axis displays descriptions for some points (eg. 100, 80, 60, 40, 20, 0) if applicable. If the graph shows 59 and 54, there may be no Y-axis description for those values.
You may think that the graph is automatically zoomed to display the currently selected values.

That was what I expected, but then there was also y-description on 61 and 59...

hdsentinel wrote:I just noticed that you mentioned that you started the test with the WD tool, NOT with Hard Disk Sentinel.
So you should not be surprised if the software was confused, if you did not use it to start the test?
The disk reports for the software that the "test is running" or "test is completed with/without error", but not WHICH test was started.

Well, the WD tool has the advantage of fixing bad sectors that were found during the test (at least, it says so ). And of course a manufacturer wants a test report of its own tool for RMA reasons.

There is some great linux tool called GSMartControl (http://gsmartcontrol.berlios.de/home/in ... creenshots) that is used by for example the Parted Magic collection. It can display self test logs...I haven't used this feature on my WD drive as I only had the need in using partitioning software after restoring my data on my current drive, but even on the much older Hitachi there are plenty of tests logged AND you can distinguish between short and long ones. Do you think this is also just estimated based on how long these tests have been run?

hdsentinel wrote:Again, Please use Hard Disk Sentinel for starting the tests because it provides more information about the results and logs the results so it can be used for future reference.

Sorry, registered users only

Bzzz wrote:That was what I expected, but then there was also y-description on 61 and 59...

I see. I'm checking

Bzzz wrote:Well, the WD tool has the advantage of fixing bad sectors that were found during the test (at least, it says so ).

Not surprising Hard Disk Sentinel has all tests included in the WD tool and even more as it constantly monitors changes of attributes during the tests. That's why it is much more sophisticated than the tool of any manufacturers - and it works on all drives, not only on the drives for that specific manufacturer.

Bzzz wrote:And of course a manufacturer wants a test report of its own tool for RMA reasons.

Yes, it is true.

Bzzz wrote:There is some great linux tool called GSMartControl...

Yes I know.
There are several such logs available on the disks but there are numerous problems with them.
First of all, they are not general and not available on all disks. So if you can check it on one or two disks, you're lucky. Hitachi disks are (a bit) better in this area because the former owner (IBM) developed this structure so IBM and Hitachi drives may store and report this information (in some cases) correctly. However, in most cases this information is not really helpful.

An other problem is that all software must communicate with the HDDs by using the hard disk controller and its driver. These drivers are in most cases incomplete and blocks several commands. This includes the detection of the disk logs and the use of many other disk related functions. For example you may try to use the acoustic management feature (Disk -> Acoustic Management Configuration, Free Fall Control, Advanced Power Management control) to check if your driver supports or blocks those commands. Even if they are supported, there is no warranty that software may detect this information and the logs contain really useful information.

Anyway, we're constantly researching about possible ways to detect and display further (useful) information.

Hi again

I just noticed that when having 50 entries in your log, there are two glitches:

* If another warning is added, the oldest one is not displayed anymore (list is limited to 50 entries)

* Any _new_ warning will not be visible in the log until HD Sentinel is sent to tray and restored or a manual update poll is made. However, they are shown in the overview tab (well, if the warning is about things that are displayed in the overview, of course ). Switching between the tabs does not fix it, you really have to take one of these actions to update the log.

Bye

Hi! Thanks for the new additions

Bzzz wrote:* If another warning is added, the oldest one is not displayed anymore (list is limited to 50 entries)

Yes, it's true. As you may read in the Help (Components of the application - Main menu - Physical disk information - Log)
"The current version of the log stores the most recent 50 problems only. After this, the oldest problems are discarded."
This may be changed in a later version.

Bzzz wrote:* Any _new_ warning will not be visible in the log until HD Sentinel is sent to tray and restored or a manual update poll is made. However, they are shown in the overview tab (well, if the warning is about things that are displayed in the overview, of course ). Switching between the tabs does not fix it, you really have to take one of these actions to update the log.

Thanks for your attention, I'm checking this!