Massive errors with new HDS launch and Pagefile.sys

[alan-0000]

I moved Pagefile.sys from my system OCZ SSD to my secondary WDC HDD E:\
I changed the HDS launch from "Load with Windows" to "Use as Service" and tried that but did not like it.
I changed the HDS launch back to "Load with Windows".
Question - has that completely undone all changes introduced by "Use as Service" ?

Big problems have arisen :-
My WDC HDD is now as it always has been, Disk 0
Until today's changes :
My Samsung HDD was normally Disk 1
and my OCZ SSD was normally Disk 2
but sometimes they swapped numbers.
Since these changes
My Samsung HDD is normally Disk 2
and my OCZ SSD is normally Disk 1

Is it possible these changes are due to the location of the 8 GB Pagefile affecting Disk numbers which I believe are allocated by the BIOS before Windows is launched ?

I have today sent several Developer test reports.
I have run the "Short Self Test" and "Random Seek Test" on the SSD and both HDD with no apparent errors.

Bigger problem
Now with a "normal" start of OCZ SSD = Disk 1
I typically see that HDS Gui opens and shows 13,027,621 raw data (SMART attrib 1) when the SSD has only Read 7 MB and Written 1 MB.
Until today the SSD had zero raw data until 100 MB was read and then it only leapt to a few thousand,
Now I am seeing Raw Data that is comparable to what I previously saw when reading all 11 GB of my SSD.
Perhaps ECC is still correcting the errors, but it has to work 10,000 times harder than before.

N.B.
Immediately before altering the HDS launch I changed the Pagefile.sys location,
and upon start-up I observed 0 MB Read and 0 MB written to the WDC so the new pagefile on WDC gets very little traffic.
I think 8 MB of RAM is more than enough for what I do.

Any advice would be appreciated.

Regards
Alan

[hdsentinel]

Dear Alan,

> I changed the HDS launch from "Load with Windows" to "Use as Service" and tried that but did not like it.
> I changed the HDS launch back to "Load with Windows".
> Question - has that completely undone all changes introduced by "Use as Service" ?

If you switched back to application mode by using the shortcut "Stop Service" in the Hard Disk Sentinel program group and answering to the question that "do you want to switch back..." then yes of course, the software is switched back to application mode. Then it can load with Windows exactly as before.

In service mode, the user interface under Windows 7 is not accessible due to the restriction added by Microsoft as only NT4, 2000, XP, 2003 allowed the use of "interactive services". This is described in the Help.
However, the latest beta versions offers limited functionality (to access hard disk tray icons and hard disk status information) even in Service mode.
The Service mode is mostly designed for systems without active user, eg. servers where the software was configured to issue alerts and then works in service mode.


> Big problems have arisen :-
> My WDC HDD is now as it always has been, Disk 0
> Until today's changes :
> My Samsung HDD was normally Disk 1
> and my OCZ SSD was normally Disk 2
> but sometimes they swapped numbers.
> Since these changes
> My Samsung HDD is normally Disk 2
> and my OCZ SSD is normally Disk 1

It is interesting why Windows changes the numbering of the drives.
What happens if you manually exchange the connections (drive positions) on the motherboard? Not sure if help, but may worth a try.

> Is it possible these changes are due to the location of the 8 GB Pagefile affecting Disk numbers which I believe are
> allocated by the BIOS before Windows is launched ?

No. The change of the swapfile does not cause that the drive numbers change.
Also no: the order applied in BIOS may not neccessarily the same as in Windows. In most cases it is the same, but this may not happen always.


> I have today sent several Developer test reports.
> I have run the "Short Self Test" and "Random Seek Test" on the SSD and both HDD with no apparent errors.

Thanks for the report, I checked and see no problems at all, Hard Disk Sentinel shows no issues, as I see both HDDs and the SSD is perfect.


> I typically see that HDS Gui opens and shows 13,027,621 raw data (SMART attrib 1) when the SSD has only Read 7 MB and Written 1 MB.
> Until today the SSD had zero raw data until 100 MB was read and then it only leapt to a few thousand,
> Now I am seeing Raw Data that is comparable to what I previously saw when reading all 11 GB of my SSD.
> Perhaps ECC is still correcting the errors, but it has to work 10,000 times harder than before.

In different forum topics we previously discsussed that these numbers are NOT errors.
According the SF specifications, the numbers are the total read operations.
The higher number only means that the SSD has more completed read operations, but this number does not indicate any kind of errors.


> Immediately before altering the HDS launch I changed the Pagefile.sys location,
> and upon start-up I observed 0 MB Read and 0 MB written to the WDC so the new pagefile on WDC gets very little traffic.
> I think 8 MB of RAM is more than enough for what I do.

8 GB RAM I suspect
Did you configure the swapfile to be FIXED size? (the minimum and maximum size should be same)
If so, I agree that the swapfile utilisation will be minimal. The default Windows setting (automatically adjust swapfile size) is a real pain, may cause high disk utilisation.

[hdsentinel]

Here are two screenshots about the S.M.A.R.T. of an ADATA S599 SSD which has same SF 12xx controller as yours (just it has temperature sensor also):





As you may see, the Data column of the "1 raw read error rate" attribute shows a huge increase, caused by a Disk -> Surface test -> Read test in Hard Disk Sentinel. The Data column reflects the high number of read operations.
For this attribute, the "Value" field and its relation to the "Threshold" may indicate problems, if the "Value" would drop below the threshold, it would suggest problem with this attribute.

[alan-0000]

Thanks

I got quite nervous about the thousand fold increase which was suddenly occurring,
especially since my nearly new SSD is the replacement for an SSD that catastrophically failed when almost new.

I still think of Raw Read Error Rate as meaning bad things because :-
It has the word "Error" ;
and I have two off HDD which never show anything but 000000000 etc in their Data fields,
my SSD normally shows 000000000 until it abruptly jumps by several thousand when a slow increase in Total Data Read has increased well above 100 MB
I would have expected a linear increase of "Raw Read ..." with Total Data Read.

I suspect that Sandforce may have some special tweaks in their firmware that cause this discontinuity.

Disc number changes never happen when the computer is put to sleep - after computer wake-up the numbers remain as they were
Changes only happen for a total shutdown and reboot.
I assumed the above indicates that the BIOS makes the mistake.

It has been pointed out on the OCZ forum that I should only have 1 "active" partition on a system.
My originally Windows 7 system on the SamSung HDD is still intact including the "Aciive" flag on its own 100 MB "System Reserved" partition.
By cancelling that "active" flag I hope to achieve stable Disc numbers.

Regards
Alan

[hdsentinel]

Personally I completely understand that seeing the increase of the counter can be annoying - but this is completely normal.
As described in the Help of Hard Disk Sentinel:

For some (especially statistical) attributes, the number can be very high and then it can fall, it is normal also. The graph is important for the critical attributes (for example reallocated, bad sector count) where the analysis of the previous values can reveal problematic days or period in the lifetime of the disk.

indicating that some values can increase without indicating problems.

I can confirm that this attribute COULD show error conditions, but not this field (which shows the amount of read sectors according Sandforce specs), but the Value field which is really adjusted upon error conditions.

It may be not a good idea to compare this value with hard disk(s) or even with other SSDs.
The attribute on some hard disks (like your WD), works very differently: in normal situations it does not show increase (as you can see). In contrast, Seagate drives show increase of "Raw read error rate" exactly as for your SSD - but this also does not mean errors.

[alan-0000]

Thanks

Your answers have reassured me.

Apart from the reference to Seagate

A few years ago my son had a dead drive and Googled the phrase "bricked Seagate drive" and Google gave him
About 90,400 results (0.32 seconds).
He eventually selected a solution and then I had to help by fabricating a special serial interface to fix the firmware of his Seagate drive

I am happy to say that my SSD has not relapsed back into "spectacularly bad" behavior.

Regards
Alan

[hdsentinel]

Me thanks too

Sorry for the reference with the Seagate drive, I just wanted to point on that not only SSD (or SSD controller) manufacturers, but different hard disk manufacturers (WD vs Seagate) may use some attributes differently.
This is why it may not be good to compare strictly one value provided by an SSD or a HDD with a completely different drive.

Anyway, if I can help or assist (for example if you encounter any kind of problem, if you see degradation of health), just let me know. The best is to use Report -> Send test report to developer option - then I can immediately check the complete status and quickly compare with the previous to verify the difference and advise.

[alan-0000]

Thanks

Everything running smoothly so far

Regards
Alan