Proper HDD maintenance procedure

[tfm]

Hello,

Like many users, I've several TB of data as a result of many years of work. My whole professional life is attached to these data, and it is too much data to upload it to a cloud backup.

After a pair of alerts from Acronis True Image of a reading error from one of my HDDs, I checked the file it got stuck to and it wasn't anything important. There was no other sign of corrupted data/degradation at all. Then, I looked for the best/suggested maintenance procedures. At some point, I got to Spinrite. Then, I got to this thread:

viewtopic.php?f=1&t=28780

That clearly states that HDD Regenerator and later on thread Spinrite are no good for restoring unreadable data. Rather, they destroy already lost data forever, by writing crap or zeroing data to bad sectors and making them readable again.

However, I have also read:

http://www.recoveryforce.com/spinrite-a ... y-program/

That showed the same thing, but author ends up his analysis by stating that Spinrite is a great tool for maintenance, just not for restoring data.

Besides regular back up (which I do), I would like to know your suggestions for drive maintenance. I am focused here on the drive, i.e, how can one take best care of his drives, so lifespan is improved, data loss is prevented as earlier as possible and defects growing over time are halted/reduced when possible?

Just a note: hardware wise, I keep a double conversion ups powering my rig, air cooling on top of all HDDs, dedicated backup drives, and I disconnect the backup drives - preventing lightning damages. RAID is not an option for the moment. So I believe whatever solution is suggested, it would be 'SWish'.

[fzabkar]

At the outset you say that "my whole professional life is attached to these data", but then you finish by saying that "RAID is not an option for the moment".

Do you have WD drives? If so, here is what happens when they are hit by a power surge:

Catastrophic failures in Western Digital PCBs:
http://www.hddoracle.com/viewtopic.php? ... 119&p=5033

Although it might sound contrarian, I would "repair" any WD drive before using it for critical data. To this end I would flow a blob of solder over each of the R67 and R64 resistors. This will allow the TVS diodes to do their job and protect your data in the event of an overvoltage. If you don't do this, and if you build a RAID using WD drives alone, then you will have essentially no redundancy in the event of a PSU failure or power surge.

But that's just my opinion ...

[tfm]

At the outset you say that "my whole professional life is attached to these data", but then you finish by saying that "RAID is not an option for the moment".

You are right, it sound odd! However, financial crisis got us... Furthermore, critical data is scattered over many HDDs, not only one or a pair. So I'd have to RAID nearly all of them. And that requires quite some money, from RAID card to doubling HDDs. So the closest I got is daily automatic backup, and disconnecting power plug avoiding lightning issues.

Do you have WD drives? If so, here is what happens when they are hit by a power surge:

Catastrophic failures in Western Digital PCBs:
http://www.hddoracle.com/viewtopic.php? ... 119&p=5033

Although it might sound contrarian, I would "repair" any WD drive before using it for critical data. To this end I would flow a blob of solder over each of the R67 and R64 resistors. This will allow the TVS diodes to do their job and protect your data in the event of an overvoltage. If you don't do this, and if you build a RAID using WD drives alone, then you will have essentially no redundancy in the event of a PSU failure or power surge.

But that's just my opinion ...

Yes, I have a bunch of them! Specially WD Black. I wasn't aware of that! I'll sure dig into that. Power wise, double conversion UPS + a reliable power supply (I use PC&C Silencer 910) minimizes electrical impacts. But they are just reliable as electronics can be! So I'll check that on WD HDDs, thanks for that.

However, I was asking first advise on further SW routine for checking, looking for errors, making drive read/write throughout. I mean, is the best practice on drive maintenance nothing but trusting internal HDD management and regular use?

Also, do you experts recommend any other SW for monitoring SMART than the manufacturer SW? Any reliable SW known for interpreting SMART data fine and displays/alerts at first sign of poor health?

[fzabkar]

However, I was asking first advise on further SW routine for checking, looking for errors, making drive read/write throughout. I mean, is the best practice on drive maintenance nothing but trusting internal HDD management and regular use?

Also, do you experts recommend any other SW for monitoring SMART than the manufacturer SW? Any reliable SW known for interpreting SMART data fine and displays/alerts at first sign of poor health?

I'm no expert, but a full surface scan in real DOS with a tool such as MHDD will identify any "slow" sectors, ie those that are difficult to read.

As for SMART monitoring, manufacturer's tools such as SeaTools and Data LifeGuard are essentially useless. SeaTools reports nothing more than PASS or FAIL, meaning that a drive with 2500 bad sectors may pass while a drive with 2501 bads will fail. DLG reports the normalised values of each attribute, but doesn't report the raw values, so a drive with a normalised score of 100 might in fact have 5 reallocated sectors. Instead I would use tools such as CrystalDiskInfo or smartmontools.

[labtech]

Besides regular back up (which I do), I would like to know your suggestions for drive maintenance. I am focused here on the drive, i.e, how can one take best care of his drives, so lifespan is improved, data loss is prevented as earlier as possible and defects growing over time are halted/reduced when possible?

Other than implementing some redundancy via RAID, everything you are doing is about right. The drives will fail sooner or later. Can't prevent this from happening.

Are you asking if there is some sort of automatic software that will pop up with a warning telling "hey, bla bla drive is crapping out!"
For single drives, there isn't anything out there that I am aware off. But you can definitely check manually with software tools like crystaldisk, hddscan, smartmontools, etc.

With hardware RAID implementation, if the design is good, then the box should at least warn via sound and/or lights that a drive has failed, so that is nice.

All in all, multiple copies of the data, in whatever form, is the way to go.

[tfm]

Sorry for the late reply, guys. Thank you for all that, I've been learning a lot here -- but I still have no answer. Let me insist a bit:

Are you saying that there is no routine that the user can take that detects/prevents errors earlier other than simply using the HDD? For example, running MHDD, chkdsk, etc. from times to times?

And what if I get an error when I try to read a file, which SW does a better job, IYO?

I'm no expert, but a full surface scan in real DOS with a tool such as MHDD will identify any "slow" sectors, ie those that are difficult to read.

So do you think running a full surface scan under DOS is any different than using chkdsk or other surface scan under Win? Also, do you consider MHDD any better than other DOS scanning SW?

As for SMART monitoring, manufacturer's tools such as SeaTools and Data LifeGuard are essentially useless. SeaTools reports nothing more than PASS or FAIL, meaning that a drive with 2500 bad sectors may pass while a drive with 2501 bads will fail. DLG reports the normalised values of each attribute, but doesn't report the raw values, so a drive with a normalised score of 100 might in fact have 5 reallocated sectors. Instead I would use tools such as CrystalDiskInfo or smartmontools.

I agree manufacturer's tools don't provide any other information. But I've read that each manufacturer can interpret better the results of their own products than general SW - that seems to be a reason to run them as well, don't you think?

[fzabkar]

MHDD identifies "slow" sectors whereas other tools (eg SeaTools) are only interested in whether a sector is readable, irrespective of how long it takes.

CHKDSK can test file integrity but should not be allowed to "repair" a drive with an extensive bad sector problem. DOS scanning software is more accurate than Windows software since there are no background tasks competing for the attention of the CPU and potentially interfering with the test.

SeaTools is essentially worthless as a SMART reporting tool. It merely reports PASS or FAIL, and this is based solely on whether any critical SMART attribute value has dropped below its threshold. SeaTools does not report the raw values of the SMART attributes. This means that a drive with 2500 reallocated sectors could receive a PASS on one day and a FAIL on the following day when that same attribute records 2501 reallocated sectors. SeaTools gives no warning that the drive has degraded.

[tfm]

MHDD identifies "slow" sectors whereas other tools (eg SeaTools) are only interested in whether a sector is readable, irrespective of how long it takes.

CHKDSK can test file integrity but should not be allowed to "repair" a drive with an extensive bad sector problem. DOS scanning software is more accurate than Windows software since there are no background tasks competing for the attention of the CPU and potentially interfering with the test.

SeaTools is essentially worthless as a SMART reporting tool. It merely reports PASS or FAIL, and this is based solely on whether any critical SMART attribute value has dropped below its threshold. SeaTools does not report the raw values of the SMART attributes. This means that a drive with 2500 reallocated sectors could receive a PASS on one day and a FAIL on the following day when that same attribute records 2501 reallocated sectors. SeaTools gives no warning that the drive has degraded.

I see. Thank you!