Drive Health Software

[Syntho]

I had some drives fail in my NAS recently and it got me thinking... I’ve never run SMART tests or basically anything else on any of my other drives like I have with my NAS drives.

I’m running SCSI drives on my 9600 and I’m not quite sure how to test them. I’ve got TechTool Pro 3 and it has a surface scan feature as well as some other drives tests.

Dumb question, but what is the difference between a SMART test and a surface scan? I’m not seeing anything about SMART tests on OS9 and prior drives. Is that basically what a surface scan is? Maybe that’s all I need to ensure the health of my SCSI drives.

I also wonder how to test an SDD in that configuration. Would probably have to pop it into a G4 at least for that.

[teroyk]

I’m not seeing anything about SMART tests on OS9 and prior drives. Is that basically what a surface scan is?

Basically SMART is drive internal logic notice that it has had any problems to read or write or too big shock.
Silverlining Pro has info page where it says is smart ok, if drive support it.
Surface scan really test reading and thas why it is better.

[ssp3]

I’m running SCSI drives on my 9600 and I’m not quite sure how to test them.

FWB SCSI Configure and FWB HDT. Reading their manual is highly recommended.

[IIO]

rumor has it, that you should not use SSDs in OS9 for years without attaching them to snowleo or higher 2 times per years to check data integrity because "OS9 wouldnt support s.m.a.r.t. status".

but what could go wrong with a cell / block when you dont write to it? SSDs normally only wear out when you write to it.*)

i never really understood how smart works in practice and what it does mean to us in OS9 esp. in conjunction with the usual cheapo SSDs (with both, overprovisioning space or not)

it would be great if somebody who knows would write a short essay about "best practice" on that topic.


*)
which is why it can make sense to keep one of many internal drives a HD for heavy writing jobs such as unstuffing, sorting, or certain programming laguages where you always overwrite thousands of little files with new versions of the same file.

[Syntho]

The method that I think I worked out is to pop out my SDDs and put them into a G4 with Tiger and use TechTool Pro 4. In fact, I did just that an hour ago but TTP doesn't really do a SMART scan, it only looks at the previous SMART data the drive itself has. And even then, I'm not sure it's reading it right.

I think the best I can do, is probably attach my SSDs to a newer Mac computer (or Windows) and run a SMART scan using modern software. For my SCSI drives though? TTP can do surface scans and you can choose to just read or both read+write. There's a slight difference in a SMART test vs. a surface scan but I'm still unclear on it exactly. A SMART scan only reads from sectors, as to where a surface scan can both read and write, so I'm wondering what the real difference is between a surface scan with read only and a SMART scan. Surface scans don't apply to SSDs, though.

[ssp3]

In OSX there's DriveDX (10.6.8 and up):
https://binaryfruit.com/drivedx/help

and older versions of Smart Utility with PPC code in it:
https://www.volitans-software.com/apps/smart-utility/

Both of them, btw, rely on open source "smartmontools" command line utilities. Do a search, maybe someone has compiled them for Tiger.
https://www.smartmontools.org/wiki/Download#InstalltheOSXDarwinpackage

In OS9 FWB mentions that they do SMART, if the drive supports it. I never tried it, though. For SCSI drives there's full surface scan and option to show grown defects. See below.

[Syntho]

I think I’ll check it out then. Thanks 

[IIO]

In OS9 FWB mentions that they do SMART, if the drive supports it.

oh FWB can do it? that would be nice. i would not have tried that at all (and i dont have it installed since OS 7.1)

[IIO]

so it basically reads all files (or sectors) and the disk does the rest on its own, like it does during normal operation just for ALL files, do i understand that right?

[Syntho]

Here’s the thing that I’m confused on. A smart scan reads the sectors. A surface scan does too, but you have e option of reading + writing with a surface scan. The SMART thing is something built into the drives themselves and the HD retains the results whenever it encounters a bad sector, where a surface scan is only done manually by 3rd party software.

This leads me to ask about the difference between a manual smart scan and a read-only surface scan. They seem similar. From what I can tell when the SMART thing started making rounds it was only data that the drive encountered in everyday operation, and a surface scan was the only real way to check a drive fully. But these days there is a manual smart scan thing which probably is the same thing as a surface scan.

All of the above is just a guess. We need someone that knows what they’re doing to correct us.

[IIO]

hehe, i bet FWB wont read 2 TB volumes, but we will see soon.

and yes, the big question is what it will be worth to try to collect data about the redability or usage amount an a cell in an SSD, when the operating system´s BIOS (or a third party utility) is not able to interpret this data on startup anyway.

if you are supposed to look at the list which sectors soon wear out and then move the affected files accordingly, then is is a non-solution.

in this case it would be easier to just copy all your data one time per year.

[IIO]

that´s what it currently analyses.

"number of errors uncorrectable by ECC" is interesting, but the question is how to be able to correct it otherwise.

i had broken HDs and CD media where it was possible to copy a broken file from only after the 15th attempt, if you know what i mean.


02    0x02 Throughput Performance    
03    0x03 Spin Up Time    
04    0x04 Start/Stop Count    
05    0x05 Reallocated Sectors Count    
07    0x07 Seek Error Rate
09    0x09 Power On Hours Count
10    0x0A Spin Retry Count
12    0x0C Power Cycle Count    
184    0xB8 End-To-End error    
187    0xBB Reported (ECC-)uncorrectable Error
188    0xBC Command Timeout
193    0xC1 Load(/Unload) Cycle Count
194    0xC2 Drive Temperaturer          
195    0xC3 Hardware ECC Recovered
196    0xC4 Reallocation Event Count
197    0xC5 Current Pending Sector Count
198    0xC6 Uncorrectable Sector Count
199    0xC7 Ultra DMA CRC Error Count
201    0xC9 Soft Read Error Rate

[refinery]

honestly in my experience working in several data centers, SMART is really useless for SSDs. If its gonna die, its gonna die, usually without any warning.
You're better off using whatever utility software is provided by the manufacturer of the SSD to check drive health, which usually means popping it in a windows machine. Samsung drives wont even properly report proper wear leveling in SMART data, its literally just a linear percentage based off the total data written and not how it was written to the drive, and 100TB of large media files being written is way, way, way different use than say 100TB of database usage. You cant get the actual health of the drive unless you use Magican.
I mean, if i was going to go to the trouble of pulling the drive out of a machine, I'd want it tested thoroughly.