And multi zone error rate. This 3 values sometimes increase in my parity disk.

 

They are still low ( 10 and 20). Should i be worried about it? Is it normal? Or should just be worried if ever reach like 10.000?

 

I have screenshots from unmenu but i dont know how post it

 

Thanks in advance

 

Other than that. everything seems to be working great. Have 3 data disk /cache / parity for the past month and a half with no problem.

 

Im running UnRaid 4.5.6

 

No one?

If the numbers continually increase (un-readable sectors, and sectors pending re-allocation) you'll want to RMA the drive.

 

Basically, if there are 10 sectors pending re-allocation there are 10 un-readable sectors on the parity disk.  It is waiting for them to be written so it can re-allocate them in a spare sector. (at that time it will know what the sector should contain, since it is being written)

 

It also means if you have a data disk failure right now, there will be 10 sectors it will not be able to re-construct properly.  They might be sectors used by files, they might be sectors un-allocated on the data drive.  Do you feel lucky?  If it is a 1TB drive there will be roughly 2000000000  sectors.  Odds are good your precious and critical files will not have one of the un-readable sectors... but what if you are not lucky....

 

I'd un-assign the parity drive,  start the array without it, then re-assign it start the array once more, and have the array completely re-write parity.  Then, perform a full parity "Check" (by pressing the check button) to read all the sectors to see if there are any more un-readable sectors.

If you are lucky all the pending re-allocation sectors will be re-allocated and no new sectors will show up when you do the parity check. 

 

If you are not lucky, RMA the drive.

 

Joe L.

Thanks JOE.

 

I did follow your suggestion and now I am running the parity check. Will be back in around 10 hours to report back.

 

Thank you

Thanks JOE.

 

I did follow your suggestion and now I am running the parity check. Will be back in around 10 hours to report back.

 

Thank you

You would first need to unassign parity/start array/stop array/re-assign parity/calc new parity before subsequently doing a "Check"

 

The first set of steps would write to the entire parity disk, allowing it to re-allocate the set the sectors pending re-allocation. The "Check" then reads it to make sure what was written is readable.

parity ceck done.

 

 

2 offline uncorrectable

5 multizone errors.

 

Should I RMA? i leave to the states tomorrow so if i have to do it, now is when.

 

pls some one.

 

thx

I had a WD20EARS that had 179 "offline_uncorrectable" errors recently.  It had me worried but then one day these errors simply disappeared. 

 

Not sure what happened so I'm watching that drive very closely.

parity ceck done.

 

 

2 offline uncorrectable

5 multizone errors.

 

Should I RMA? i leave to the states tomorrow so if i have to do it, now is when.

 

pls some one.

 

thx

If you performed the un-assign/start array/stop array/re-assign/initial parity calc/subsequent parity check sequence, and you ended up with several uncorrectable errors, then the errors are still trickling in, and I'd say you might want to RMA the drive.  It is borderline, but odds are it will not get better.

parity ceck done.

 

 

2 offline uncorrectable

5 multizone errors.

 

Should I RMA? i leave to the states tomorrow so if i have to do it, now is when.

 

pls some one.

 

 

 

thx

If you performed the un-assign/start array/stop array/re-assign/initial parity calc/subsequent parity check sequence, and you ended up with several uncorrectable errors, then the errors are still trickling in, and I'd say you might want to RMA the drive.   It is borderline, but odds are it will not get better.

 

Ok, then i will be taking the disk out tomorrow to bring it with me to the states.Guess is my only option. I would leave the server turned off for the next 2 weeks until I am back here. Any real complication on this?? or should i just get a new disk in the states and RMA later in my next trip (March)

 

Any better idea?

Well i was reading RMA policy in Newegg. Seems to be just 30 days and I have this disk for over 2 months now. So i guess im not taking it with me in my trip =(

I had a WD20EARS that had 179 "offline_uncorrectable" errors recently.  It had me worried but then one day these errors simply disappeared. 

 

Not sure what happened so I'm watching that drive very closely.

 

It shouldn't be just disappeared, "offline uncorrectable" error is similar to "reallocated sector" error, it is "offline" because those errors are detected by self-diagnostic facility in hard disk that will be triggered whenever disk is idle for certain period of time. sometime it will just randomly wandering around and reading sectors here and there.

2 offline uncorrectable

5 multizone errors.

 

 

i have a Seagate 1.5TB disk has now 51 "reallocated sector" errors and still using it.

unless you saw those errors increasing in short period of time otherwise base on my experience

hard disk will not fail suddenly. This disk had about 30/40 reallocated sectors for long time then jump to

50 mark within couple weeks.

 

I emailed Seagate tech support, i was told there are about 180 spare sectors available

for remapping however i can RMA this disk because it is still within warranty

 

more remapped sectors will degrade disk performance because remapping is similar to "indirect addressing"

but i don't think it is really noticeable.

 

 

i have a Seagate 1.5TB disk has now 51 "reallocated sector" errors and still using it.

unless you saw those errors increasing in short period of time otherwise base on my experience

hard disk will not fail suddenly. This disk had about 30/40 reallocated sectors for long time then jump to

50 mark within couple weeks.

 

This is not a good way to think about reallocated sectors.  Whether the disk is going to completely fail or not is not the issue.  In order for unRAID to rebuild a disk, every sector on every other disk must be 100% readable.  If you have a disk that is slowly reallocating sectors, even if only a few a month, chances are that if a different disk fails (or you want to upsize a different disk), a couple of bad sectors (needing reallocation) will be encountered - and the drive may not be able to read them.  unRAID will not stop the reconstruction, but you may be left with a few badly reconstructed sectors on the rebuilt disk WITH NO WAY TO FIGURE OUT WHAT FILES ARE CORRUPTED.  And unless you are pretty knowledable, you might not even know that this corruption had happened.  This is the Achilles heal of unRAID IMO.

 

In order for this not to happen, you have to be able to trust EVERY sector on EVERY disk in your array.  Disks can't be reporting new pending or reallocated sectors.  Period.

 

In my experience personally and having consistantly participated in these forums for 3 years, most every disk has zero reallocated sectors.  VERY occasionally 1 or 2 will develop during the break in period and hold steady.  But once a broken in drive starts reporting reallocated sectors - even ONE - the drive is going to keep reallocating sectors.  Of course there are rare exceptions, but this is the general rule.

 

My advice - once you start seeing reallocated sectors - keep running parity checks.  If you can run three in a row without a new reallocated/pending sector, you might feel somewhat safe that the reallocated sectors will hold steady.  But I've never seen anyone able to run 3 consective parity checks and not continue to get reallocations or pending reallocations.  So why do I suggest doing it?  Two reasons:

 

1 - to convince the person the disk really is failing

 

2 - to get evidence for RMA or return the disk.  (Reporting 1 reallocated sector is not going to be accepted as a broken disk, but being able to show that every time you read the disk more sectors are reallocating IS good evidence.)

 

I emailed Seagate tech support, i was told there are about 180 spare sectors available

for remapping however i can RMA this disk because it is still within warranty

 

Common understanding was that there are several thousand spare sectors.  I've seen smart reports showing wll over a 1000 reallocated sectors.  Maybe some disks models only have 180 - but in general I think that number is way low.  But whether there are 180 or 18,000 spare sectors, doesn't matter.  It matters if your drive keeps using them.

 

more remapped sectors will degrade disk performance because remapping is similar to "indirect addressing"

but i don't think it is really noticeable.

 

Yes, the remapped sectors would not be physically contiguous with the other disk sectors.  But the performance lost with a couple of these on a disk is not a big deal.  The big deal is not being able to trust every sector on the disk should it be needed for disk reconstruction.

In order for this not to happen, you have to be able to trust EVERY sector on EVERY disk in your array.  Disks can't be reporting new pending or reallocated sectors.  Period.

 

The question is we will never know when/how a bad sector will happen, the one thing we as well as disk vendor know is bad sector will happen. that is why there is a pool of spare sectors. If user really worry about data integrity then s/he need to use other way to protect it such as CRC or MD5, this is same as in network communication, line generally is reliable but to make sure data is not corrupted from A to B, use CRC to protect it.

 

My advice - once you start seeing reallocated sectors - keep running parity checks.  If you can run three in a row without a new reallocated/pending sector, you might feel somewhat safe that the reallocated sectors will hold steady.  But I've never seen anyone able to run 3 consective parity checks and not continue to get reallocations or pending reallocations.  So why do I suggest doing it?  Two reasons:

 

1 - to convince the person the disk really is failing

 

2 - to get evidence for RMA or return the disk.  (Reporting 1 reallocated sector is not going to be accepted as a broken disk, but being able to show that every time you read the disk more sectors are reallocating IS good evidence.)

 

If you keep doing parity check you might have more chance to find bad sectors, the question is how many parity check is enough to declare disk is bad sector free? or should we say is there any bad-sector free disk? Take this 1.5TB segate disk i had as example, it started with about 20 something bad sectors then gradually going up to 40 over many months then stay there for long time, at least half year if my memory serves me, then up again to 51. there were many disk access, monthly parity check as well as manually parity checks because i was doing data reconstruction for other disks.

 

users still can RMA a disk because of bad sectors as i was told by Seagate Tech support, however i wonder if everyone start to RMA a disk because of handful bad sectors, i think the end result will be vendors have to raise disk price in order to compensate those cost or they have to find a way to make a better quality disk hence higher price tag? (enterprise vs consumer grad) or they just totally out of business?  . Fortunately, most of computer users don't really know about this little secret of bad sector.

 

 

The question is we will never know when/how a bad sector will happen, the one thing we as well as disk vendor know is bad sector will happen. that is why there is a pool of spare sectors. If user really worry about data integrity then s/he need to use other way to protect it such as CRC or MD5, this is same as in network communication, line generally is reliable but to make sure data is not corrupted from A to B, use CRC to protect it.

 

I agree.  I actually prefer .PAR2 sets.  I wish something like that were integral in unRAID.

 

If you keep doing parity check you might have more chance to find bad sectors, the question is how many parity check is enough to declare disk is bad sector free? or should we say is there any bad-sector free disk? Take this 1.5TB segate disk i had as example, it started with about 20 something bad sectors then gradually going up to 40 over many months then stay there for long time, at least half year if my memory serves me, then up again to 51. there were many disk access, monthly parity check as well as manually parity checks because i was doing data reconstruction for other disks.

 

If a disk makes it through a preclear cycle and monthly parity checks with no reallocated sectors, I am happy.  My experience has been that most disks will stay reallocated sector free through their useful life (3-5 years).

 

I refer to a disk that starts getting reallocated sectors as "unraveling".  Its like the first one creates a little string.  And every read through the disk tugs on the string unraveling the disk a little more.  Sooner or later so much of the disk is unraveled that it is useless.  When I see this start to happen, I'm done with that disk.

 

But if you like to live on the edge , realize that just because a disk has reallocated sectors doesn't mean a sector actually failed.  The disk can detect marginal sectors before they get so bad and become completely unreadable.  If a sector is marginal, the disk can reallocate it directly.  So not every reallocated sector would have lead to a corrupted rebuild, just some of them.

 

users still can RMA a disk because of bad sectors as i was told by Seagate Tech support, however i wonder if everyone start to RMA a disk because of handful bad sectors, i think the end result will be vendors have to raise disk price in order to compensate those cost or they have to find a way to make a better quality disk hence higher price tag? (enterprise vs consumer grad) or they just totally out of business?  . Fortunately, most of computer users don't really know about this little secret of bad sector.

 

If a disk starts unraveling, I am going to replace it.  If that makes the price of disks go up, so be it.  But I have had phenominally good luck with drives and never had one of my own unravel.  My array has no reallocated sectors on any disks.