Jump to content

What does this mean, please?


bobrap

Recommended Posts

TBH, I wouldn't worry so much about 2 reallocated sectors.  When you start start worrying is when they start to increase.  I have numerous drives with 1-2 reallocated, and they never increase.  (For that matter, I have one that has 367 reallocated, and it hasn't changed in 5 years)

 

Some people feel that any reallocated sector is a sign of impending doom.  Not necessarily.  Its when they begin to grow after the first few develop that's the sign of doom.  But, then again I'm from the era when every single ST506 hard drive came with a label right on them with what the bad sectors were.  Back then you didn't trust a drive if it DIDN'T have at least one bad sector on it.

Link to comment

Sorry to be the bearer of bad news, but drive 5 is failing. SMART is a test run on the harddrive that checks it for signs of an upcoming death. I'd recommend getting a new drive in there ASAP.

 

Disk 5 is fine, 2 reallocated sectors after 31300 operational hours is fairly normal.  Most important is that the VALUE and WORST have not dropped from 100, essentially perfect.  And if you have older SMART reports for this drive and can determine that those 2 sectors were reallocated well before now, then you should be OK.  If they are recent, then the drive does need to be thoroughly tested.

 

I personally believe all drives have marginal or bad sectors, but they ship it with the best tracks selected and the rest as spares.  Then they set the initial count of reallocated sectors to zero.  You have many more spare sectors to use.  The problem is that if you see this number increasing, then there are probably mechanical issues with the drive, and that's serious.  Our advice is always to watch the reallocated number, and if it isn't changing, you are OK.  If it IS increasing, then test it hard, make it fail if you can.  If it stops increasing no matter how hard you test it, then the marginal sectors have been isolated and replaced, and you're OK again.

 

The display of SMART info is new, still being worked on.  A non-zero value is often not an issue, so long as it is not changing.  The triggering has been improved to look for changes, not non-zero values, but it also needs to consider the first look as not a change either, even if non-zero.

Link to comment

Sorry to be the bearer of bad news, but drive 5 is failing. SMART is a test run on the harddrive that checks it for signs of an upcoming death. I'd recommend getting a new drive in there ASAP.

 

Disk 5 is fine, 2 reallocated sectors after 31300 operational hours is fairly normal.  Most important is that the VALUE and WORST have not dropped from 100, essentially perfect.  And if you have older SMART reports for this drive and can determine that those 2 sectors were reallocated well before now, then you should be OK.  If they are recent, then the drive does need to be thoroughly tested.

 

2 reallocated sectors MAY not be the end of the drive (but then again it may), I would not say that it is normal after 3.5 years. Although normalized values are all we have to go on with some attributes, for reallocated sectors they tend to be a very poor indicator.

 

My experience is that even a single reallocated sector is typically a harbinger of doom for the drive. If I ever get a pending or reallocated sector, I will start running parity checks, checking the reallocation counts after each one. If it stays constant for 3 straight parity checks I'd call it "ok for now" and not replace it, but continue to look for signs of more degradation in the future. But if after 5 or 6 parity checks I still can't go three in a row without reallocations increasing, I will replace the drive. Maybe not a good analogy, but I think of it like a pothole. Once a little pavement breaks away (single reallocation), every car that drives through (disk access nearby the prior reallocation) makes it worse (causes additional reallocations). Until the pothole is big enough to swallow a car.

 

I personally believe all drives have marginal or bad sectors, but they ship it with the best tracks selected and the rest as spares.  Then they set the initial count of reallocated sectors to zero.  You have many more spare sectors to use. 

 

I don't believe all drives have marginal or bad sectors. I have had very few drives develop bad sectors even after 5+ years of use. But some drive manufacturers/models have poor reliability stats, and perhaps these drive issues are due to poor manufacturing of the platters or other design issues with those drive models.

 

As far as testing each drive, it would take a huge amount of time and effort to identify the marginal sectors. And the sensitivity of the equipment necessary to rate each sector would be pretty amazing. Non-contiguous sectors also create performance issues, and using badblocks it is possible to mark all of the reallocated sectors as "bad sectors" in the drive's format. Thus these sectors are skipped, minimally impacting sequential performance, rather than causing a large head movement and rotational delay to access the reallocated sector on a totally different part of the drive surface. If the drive manufacturers were playing games with the sectoring on the drive, this would not work very well.

 

The problem is that if you see this number increasing, then there are probably mechanical issues with the drive, and that's serious.  Our advice is always to watch the reallocated number, and if it isn't changing, you are OK.  If it IS increasing, then test it hard, make it fail if you can.  If it stops increasing no matter how hard you test it, then the marginal sectors have been isolated and replaced, and you're OK again.

 

My procedure above (three parity checks in a row) works well for this. It is VERY VERY rare that I find a disk will pass the three parity check test once it develops even a single reallocated sector develops.

 

The display of SMART info is new, still being worked on.  A non-zero value is often not an issue, so long as it is not changing.  The triggering has been improved to look for changes, not non-zero values, but it also needs to consider the first look as not a change either, even if non-zero.

 

New is the dynamix GUI, but myMain has been doing this for years. Let's you set a threshold before reporting issues a second time.

Link to comment

@bjp999

 

I got most of that, somethings went right over my head.  Your suggestion is to run a few parity checks and see if the count increases.  If it does, replace disc.  But, you said a disc with reallocated sectors rarely passes that test and that makes me think "What's the point?".  So, since I'm kinda hardware ignorant, I came here to tap some of the knowledge available.  Anyway I can run those tests quickly to see if the drive is trashed?  Right now there hasn't been a parity check done on the 14 discs because of a new config.  Should I run bad blocks, and if so, where's a good place to find out how to correctly use it.  Obviously, I need someone holding my hand. :)  Lastly, any recommendations for brands if I have to replace the drive?  Thanks for the help and advice!

Link to comment

@bjp999

 

I got most of that, somethings went right over my head.  Your suggestion is to run a few parity checks and see if the count increases.  If it does, replace disc.  But, you said a disc with reallocated sectors rarely passes that test and that makes me think "What's the point?".  So, since I'm kinda hardware ignorant, I came here to tap some of the knowledge available.  Anyway I can run those tests quickly to see if the drive is trashed?  Right now there hasn't been a parity check done on the 14 discs because of a new config.  Should I run bad blocks, and if so, where's a good place to find out how to correctly use it.  Obviously, I need someone holding my hand. :)  Lastly, any recommendations for brands if I have to replace the drive?  Thanks for the help and advice!

 

What's the point? Drives are expensive and having a few reallocated sectors is not a big deal if they are not continuing to increase. My comments are purely based on experience not theory. Tomorrow a new drive could come out that does not have the "pothole" effect I describe. So I would definitely try it. All it takes is a little time. Throwing out a drive with 1 reallocated sector seems harsh. But throwing out one that has had the reallocated sectors go from 1 to 2 to 2 to 8 to 10 to ... It is not trustworthy.

 

I would not run bad blocks on an array disk. Parity checks work nicely IMO. unRAID does exactly the right things to try to repair the reallocated sectors without corrupting data. If the drive is outside of the array, and contains no valuable data, bad blocks or preclear are options.

 

Look at the BackBlaze studies for disk recommendations. I love the HGST's, but of late the prices have gotten rather high. It is a balancing act of price, performance and reliability. Just steer clear of the drives with really bad reliability stats.

Link to comment

Thanks again.  My what's the point comment was a bit defeatist.  So, the recommendation is run the parity checks and see how it goes?  There is no data on the disc because thought it woukd be trashed.  Preclear, bad blocks or parity check?  Really appreciate the help and patience.

Link to comment

Thanks again.  My what's the point comment was a bit defeatist.  So, the recommendation is run the parity checks and see how it goes?  There is no data on the disc because thought it woukd be trashed.  Preclear, bad blocks or parity check?  Really appreciate the help and patience.

 

Is it in the array? If so, parity check.

 

If not, preclear.

 

That would be my suggestion. I have no experience with bad blocks and not sure it does the right combination of reads and writes to achieve the desired goals. Preclear is probably better than a parity check, but once a drive is in the array, a preclear it not possible.

 

Here to help. No problem with questions.

 

 

Link to comment

Thanks again.  My what's the point comment was a bit defeatist.  So, the recommendation is run the parity checks and see how it goes?  There is no data on the disc because thought it woukd be trashed.  Preclear, bad blocks or parity check?  Really appreciate the help and patience.

 

Is it in the array? If so, parity check.

 

If not, preclear.

 

That would be my suggestion. I have no experience with bad blocks and not sure it does the right combination of reads and writes to achieve the desired goals. Preclear is probably better than a parity check, but once a drive is in the array, a preclear it not possible.

 

Here to help. No problem with questions.

However, your first screenshot says parity is invalid. You need to fix that.

 

That was a new setup and I hadn't run a new parity check yet.  Knew that would catch someone's eye.

Thanks.

Not clear whether you meant you haven't yet built parity, or you just haven't run a parity check after building parity. If you haven't yet built parity there is really no impact from doing a new config to take it out of the array, then you can preclear it.
Link to comment

I now know to keep my mouth closed when it comes to SMART issues, thanks for the correction guys :)

 

Sorry, I didn't want to come off as 'correcting', but I did want to 'educate', especially bright guys who can pass it on.  When I 'travel' around the web at times, happening on various other support sites, I am constantly appalled at the misinformation and misunderstanding of SMART info and reports, and I want to help.  I constantly see the supposed gurus of sites giving terrible advice, completely misinterpreting SMART info that a poor user has presented to them.  I've created a wiki page Understanding SMART Reports, but it's terribly incomplete, not sure I should even be pointing to it yet.  It does give some info on interpreting some of the SMART info.  I really want to get back to that and finish it, some day.

Link to comment

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...