Need help with "reallocated sector count"?
in [Storage]
|
Prev: INT13/02h Read error (AX=0100, DL=80) trying xp installation in SATA
Next: Need to create a CDFS partition on a USB key
From: Arno Wagner on 14 Nov 2007 19:45 Previously kimiraikkonen <kimiraikkonen85(a)gmail.com> wrote: > On Nov 14, 12:55 pm, Arno Wagner <m...(a)privacy.net> wrote: >> Previously kimiraikkonen <kimiraikkone...(a)gmail.com> wrote: >> > Arno, Rod (no quatiton mark :-) ), Franc, >> > Thanks you for following this topic and keeping replying. >> > So, as overall when you looked at SMART values, i want to summarize >> > what i got in that topic? >> > 1- There are 98 reallocated bad sectors but there's no unallocated as >> > a threat of data loss? Right? >> >> The 98 sectors have been moved out of the way. Not risk from them. > So, they were containing data before they've got reallocated. Right? >> > 2-ID Attribute Description Threshold Value Worst Data Status >> > 05 Reallocated Sector Count 36 98 98 98 OK: Value is normal >> > That means there are 98 reallocated sectors, but still couldn't >> > completely understand for Seagate what "threshold - 36" means? It >> > shouldn't be percantage, does it mean that i'm allowed to allocate >> > 100-36 = 64 bad ones? (frustrating) >> >> You have bad sector number x. This inceases on more bad sectors >> found. By some obscure procedure that gives value y. This decreses >> with more bad secors found. A first hypothesist is that y decreses >> by 2 for every 98 bad sectors. The asumption is that the >> initial value was 100. Then you have threshold value z. >> If y ever reaches or falls below z, then you get a bad SMART status >> for the disk. > Very frustrating but good sample if it's straight forward true. > Manufacturer have to explain their raw value and Smart calculations > than guessing. >> The current speculation is that in your particular case x and y >> have the same numerical value, purely by accident. >> >> > 3- The other SMART values are fine as stated by many programs. Right? >> >> Looked that way. >> >> > 4- A "reallocated sector count" shows the amount of reallocated / >> > replaced sectors silently while the drive is operating. When the drive >> > has a problem with sector, first it tries to replace that sectors with >> > a "spare" sector thus a "reallocated sector" statistic is updated. >> >> The reallocation count will increase on any successful >> realocation. > How much sector will my drive be allowed to reallocate? (eg: What > about for a modern 40gb drive ? ) Depends. It this is really 98 down from 100 and caused by 98 reallocation, then a linear extrapolation yields about 3000 reallocated sectors, before the SMART status goes bad. I expect that at value zero, it will be out of spare sectors, but that is pure speculation. With that it could reallocate about 4000 sectors. >> Reallocation can happen later. If it was unsuccesful so far, you >> get a "pending sector". As son as it has been reallocated, >> the pending secor attribut is decreased again. >> >> > If the sector is completely unreadable or unreallocatable, you see it >> > as "bad" marked in surface scan tools like Seatools. Did i understand >> > correct? >> >> Yes. > That's why the full / surface scan utilities are there. They report > "bad" , really "bad" sectors which can't be allocated by firmware of > HDD. Right? They report bad sectors that were not reallocated yet. The disk can allways reallocate sectors, but sometimes only with dataloss or after several tries. >> > 5- Though a drive has "reallocated sectors", if it can read every data >> > (checking with several disk reading utilites), it doesn't have a >> > unallocatable / unfixable bad sectos which are the reasons of real >> > data loss. >> >> The reallocated sectros are recognized _and_ corrected problems. >> These sectors will not cause problems in the future. Other >> sectors may go bad, but if a complete surface scan does not >> show any, all data is in sectors that are fine or at least look >> like it. In rare cases a sector can work with some data, but be >> defect with other data. Very rare with today's disks, due to >> heavy use of error correcting codes. >> >> You still may want to do a complete surface scan regularly, >> so that the disk can recognize sectors slowly going bad >> in time and can rescue the data in them. I run such a test >> every 14 days automatically, but once every 1-2 months should >> do fine. You can make that a part of your standard backup >> procedure. > I usually do it if i record important data. But still "none" bad > sectors have been reported by surface scan of Seatools. > So, can i say i don't have any bad sectors present at the moment? No. But if you have bad sectors, they cannot be detected by a read only test. And the probability of such sectors is pretty small. Arno
From: kimiraikkonen on 15 Nov 2007 03:09 On Nov 15, 2:45 am, Arno Wagner <m...(a)privacy.net> wrote: > Previously kimiraikkonen <kimiraikkone...(a)gmail.com> wrote: > > On Nov 14, 12:55 pm, Arno Wagner <m...(a)privacy.net> wrote: > >> Previously kimiraikkonen <kimiraikkone...(a)gmail.com> wrote: > >> > Arno, Rod (no quatiton mark :-) ), Franc, > >> > Thanks you for following this topic and keeping replying. > >> > So, as overall when you looked at SMART values, i want to summarize > >> > what i got in that topic? > >> > 1- There are 98 reallocated bad sectors but there's no unallocated as > >> > a threat of data loss? Right? > > >> The 98 sectors have been moved out of the way. Not risk from them. > > So, they were containing data before they've got reallocated. Right? > >> > 2-ID Attribute Description Threshold Value Worst Data Status > >> > 05 Reallocated Sector Count 36 98 98 98 OK: Value is normal > >> > That means there are 98 reallocated sectors, but still couldn't > >> > completely understand for Seagate what "threshold - 36" means? It > >> > shouldn't be percantage, does it mean that i'm allowed to allocate > >> > 100-36 = 64 bad ones? (frustrating) > > >> You have bad sector number x. This inceases on more bad sectors > >> found. By some obscure procedure that gives value y. This decreses > >> with more bad secors found. A first hypothesist is that y decreses > >> by 2 for every 98 bad sectors. The asumption is that the > >> initial value was 100. Then you have threshold value z. > >> If y ever reaches or falls below z, then you get a bad SMART status > >> for the disk. > > Very frustrating but good sample if it's straight forward true. > > Manufacturer have to explain their raw value and Smart calculations > > than guessing. > >> The current speculation is that in your particular case x and y > >> have the same numerical value, purely by accident. > > >> > 3- The other SMART values are fine as stated by many programs. Right? > > >> Looked that way. > > >> > 4- A "reallocated sector count" shows the amount of reallocated / > >> > replaced sectors silently while the drive is operating. When the drive > >> > has a problem with sector, first it tries to replace that sectors with > >> > a "spare" sector thus a "reallocated sector" statistic is updated. > > >> The reallocation count will increase on any successful > >> realocation. > > How much sector will my drive be allowed to reallocate? (eg: What > > about for a modern 40gb drive ? ) > > Depends. It this is really 98 down from 100 and caused by > 98 reallocation, then a linear extrapolation yields about > 3000 reallocated sectors, before the SMART status goes bad. > I expect that at value zero, it will be out of spare sectors, > but that is pure speculation. With that it could reallocate > about 4000 sectors. If a sector is about 512bytes, we can can't we calculate how many they are? > >> Reallocation can happen later. If it was unsuccesful so far, you > >> get a "pending sector". As son as it has been reallocated, > >> the pending secor attribut is decreased again. > > >> > If the sector is completely unreadable or unreallocatable, you see it > >> > as "bad" marked in surface scan tools like Seatools. Did i understand > >> > correct? > > >> Yes. > > That's why the full / surface scan utilities are there. They report > > "bad" , really "bad" sectors which can't be allocated by firmware of > > HDD. Right? > > They report bad sectors that were not reallocated yet. > The disk can allways reallocate sectors, but sometimes only > with dataloss or after several tries. So, what's the difference between reallocation bad sectors and zero- fill (low-level sector replacement)? At past, i remember 2 bad sectors were zero-filled(replaced) with full scan tool Seatools. Also i remember, 32-40kb bad blocks were markend as bad in Scandisk (chkdsh) automatically. But after formatting and low-leveling disk, "none" bad sectors are shown during a Dos based tool Seatool surface scan and chkdsk. > > > >> > 5- Though a drive has "reallocated sectors", if it can read every data > >> > (checking with several disk reading utilites), it doesn't have a > >> > unallocatable / unfixable bad sectos which are the reasons of real > >> > data loss. > > >> The reallocated sectros are recognized _and_ corrected problems. > >> These sectors will not cause problems in the future. Other > >> sectors may go bad, but if a complete surface scan does not > >> show any, all data is in sectors that are fine or at least look > >> like it. In rare cases a sector can work with some data, but be > >> defect with other data. Very rare with today's disks, due to > >> heavy use of error correcting codes. > > >> You still may want to do a complete surface scan regularly, > >> so that the disk can recognize sectors slowly going bad > >> in time and can rescue the data in them. I run such a test > >> every 14 days automatically, but once every 1-2 months should > >> do fine. You can make that a part of your standard backup > >> procedure. > > I usually do it if i record important data. But still "none" bad > > sectors have been reported by surface scan of Seatools. > > So, can i say i don't have any bad sectors present at the moment? > > No. But if you have bad sectors, they cannot be detected by a > read only test. And the probability of such sectors is pretty > small. Rod said different? Everybody says different things :-( Isn't there the truth? Again, rewinding the topic (seems that topic will break group record :-) ), what do you mean by saying read-only test? Why can't i see if there bad sectors? What is full surface scan tool used for then??? > Arno Thanks.
From: Rod Speed on 15 Nov 2007 03:49 kimiraikkonen <kimiraikkonen85(a)gmail.com> wrote > Arno Wagner <m...(a)privacy.net> wrote >> kimiraikkonen <kimiraikkone...(a)gmail.com> wrote >>> Arno Wagner <m...(a)privacy.net> wrote >>>> kimiraikkonen <kimiraikkone...(a)gmail.com> wrote >>>>> Arno, Rod (no quatiton mark :-) ), Franc, >>>>> Thanks you for following this topic and keeping replying. >>>>> So, as overall when you looked at SMART values, i want to >>>>> summarize >>>>> what i got in that topic? >>>>> 1- There are 98 reallocated bad sectors but there's no >>>>> unallocated as >>>>> a threat of data loss? Right? >> >>>> The 98 sectors have been moved out of the way. Not risk from them. >>> So, they were containing data before they've got reallocated. Right? >>>>> 2-ID Attribute Description Threshold Value Worst >>>>> Data Status 05 Reallocated Sector Count 36 >>>>> 98 98 98 OK: Value is normal That means there are >>>>> 98 reallocated sectors, but still couldn't >>>>> completely understand for Seagate what "threshold - 36" means? It >>>>> shouldn't be percantage, does it mean that i'm allowed to allocate >>>>> 100-36 = 64 bad ones? (frustrating) >> >>>> You have bad sector number x. This inceases on more bad sectors >>>> found. By some obscure procedure that gives value y. This decreses >>>> with more bad secors found. A first hypothesist is that y decreses >>>> by 2 for every 98 bad sectors. The asumption is that the >>>> initial value was 100. Then you have threshold value z. >>>> If y ever reaches or falls below z, then you get a bad SMART status >>>> for the disk. >>> Very frustrating but good sample if it's straight forward true. >>> Manufacturer have to explain their raw value and Smart calculations >>> than guessing. >>>> The current speculation is that in your particular case x and y >>>> have the same numerical value, purely by accident. >> >>>>> 3- The other SMART values are fine as stated by many programs. >>>>> Right? >> >>>> Looked that way. >> >>>>> 4- A "reallocated sector count" shows the amount of reallocated / >>>>> replaced sectors silently while the drive is operating. When the >>>>> drive >>>>> has a problem with sector, first it tries to replace that sectors >>>>> with >>>>> a "spare" sector thus a "reallocated sector" statistic is updated. >> >>>> The reallocation count will increase on any successful >>>> realocation. >>> How much sector will my drive be allowed to reallocate? (eg: What >>> about for a modern 40gb drive ? ) >> >> Depends. It this is really 98 down from 100 and caused by >> 98 reallocation, then a linear extrapolation yields about >> 3000 reallocated sectors, before the SMART status goes bad. >> I expect that at value zero, it will be out of spare sectors, >> but that is pure speculation. With that it could reallocate >> about 4000 sectors. > If a sector is about 512bytes, The size of the sector isnt relevant to that number. > we can can't we calculate how many they are? Because you dont know the algorithm used to produce that particular number. And in that case the problem with the english is you didnt proofread your post, the first 'we' should presumably be why. >>>> Reallocation can happen later. If it was unsuccesful so far, you >>>> get a "pending sector". As son as it has been reallocated, >>>> the pending secor attribut is decreased again. >> >>>>> If the sector is completely unreadable or unreallocatable, you >>>>> see it >>>>> as "bad" marked in surface scan tools like Seatools. Did i >>>>> understand >>>>> correct? >> >>>> Yes. >>> That's why the full / surface scan utilities are there. They report >>> "bad" , really "bad" sectors which can't be allocated by firmware of >>> HDD. Right? >> >> They report bad sectors that were not reallocated yet. >> The disk can allways reallocate sectors, but sometimes only >> with dataloss or after several tries. > So, what's the difference between reallocation bad sectors > and zero-fill (low-level sector replacement)? Not much. Zero fill is used as a write to the sector which may well see the drive decide that it can reallocate the sector because the user clearly doesnt care about the sector contents anymore. And zero fill has nothing to do with low level except that a low level format will normally just be a zero fill with modern hard drives that cant actually do a low level format anymore. > At past, In the past, > i remember 2 bad sectors were zero-filled(replaced) with full scan tool Seatools. And that allows the drive to see that you dont care about the data in those sectors. > Also i remember, 32-40kb bad blocks were markend as bad > in Scandisk (chkdsh) automatically. But after formatting and > low-leveling disk, "none" bad sectors are shown during a > Dos based tool Seatool surface scan and chkdsk. Because the low level formatting would just be zero filling with modern drives and the drive will reallocate those sectors once it realises that you dont want the data in those sectors. >>>>> 5- Though a drive has "reallocated sectors", if it can read every >>>>> data (checking with several disk reading utilites), it doesn't >>>>> have a >>>>> unallocatable / unfixable bad sectos which are the reasons of real >>>>> data loss. >> >>>> The reallocated sectros are recognized _and_ corrected problems. >>>> These sectors will not cause problems in the future. Other >>>> sectors may go bad, but if a complete surface scan does not >>>> show any, all data is in sectors that are fine or at least look >>>> like it. In rare cases a sector can work with some data, but be >>>> defect with other data. Very rare with today's disks, due to >>>> heavy use of error correcting codes. >> >>>> You still may want to do a complete surface scan regularly, >>>> so that the disk can recognize sectors slowly going bad >>>> in time and can rescue the data in them. I run such a test >>>> every 14 days automatically, but once every 1-2 months should >>>> do fine. You can make that a part of your standard backup >>>> procedure. >>> I usually do it if i record important data. But still "none" bad >>> sectors have been reported by surface scan of Seatools. >>> So, can i say i don't have any bad sectors present at the moment? >> No. But if you have bad sectors, they cannot be detected by a >> read only test. And the probability of such sectors is pretty small. > Rod said different? No I didnt. He's rather clumsily saying that some sectors can look fine on a read, but not on a write with particular data used for the write. Its not that common for the data to make a difference to the sector being bad and I didnt comment on that situation previously in this thread. > Everybody says different things :-( Thats usenet for ya |-( > Isn't there the truth? Fraid not. > Again, rewinding the topic (seems that topic will break group record :-) ), Not a chance. > what do you mean by saying read-only test? A test that only reads sectors, doesnt write to the drive. > Why can't i see if there bad sectors? You can with the SMART report. > What is full surface scan tool used for then??? Its what was used before SMART showed up.
From: Arno Wagner on 15 Nov 2007 05:13 Previously kimiraikkonen <kimiraikkonen85(a)gmail.com> wrote: > On Nov 15, 2:45 am, Arno Wagner <m...(a)privacy.net> wrote: >> Previously kimiraikkonen <kimiraikkone...(a)gmail.com> wrote: >> > On Nov 14, 12:55 pm, Arno Wagner <m...(a)privacy.net> wrote: >> >> Previously kimiraikkonen <kimiraikkone...(a)gmail.com> wrote: >> >> > Arno, Rod (no quatiton mark :-) ), Franc, >> >> > Thanks you for following this topic and keeping replying. >> >> > So, as overall when you looked at SMART values, i want to summarize >> >> > what i got in that topic? >> >> > 1- There are 98 reallocated bad sectors but there's no unallocated as >> >> > a threat of data loss? Right? >> >> >> The 98 sectors have been moved out of the way. Not risk from them. >> > So, they were containing data before they've got reallocated. Right? >> >> > 2-ID Attribute Description Threshold Value Worst Data Status >> >> > 05 Reallocated Sector Count 36 98 98 98 OK: Value is normal >> >> > That means there are 98 reallocated sectors, but still couldn't >> >> > completely understand for Seagate what "threshold - 36" means? It >> >> > shouldn't be percantage, does it mean that i'm allowed to allocate >> >> > 100-36 = 64 bad ones? (frustrating) >> >> >> You have bad sector number x. This inceases on more bad sectors >> >> found. By some obscure procedure that gives value y. This decreses >> >> with more bad secors found. A first hypothesist is that y decreses >> >> by 2 for every 98 bad sectors. The asumption is that the >> >> initial value was 100. Then you have threshold value z. >> >> If y ever reaches or falls below z, then you get a bad SMART status >> >> for the disk. >> > Very frustrating but good sample if it's straight forward true. >> > Manufacturer have to explain their raw value and Smart calculations >> > than guessing. >> >> The current speculation is that in your particular case x and y >> >> have the same numerical value, purely by accident. >> >> >> > 3- The other SMART values are fine as stated by many programs. Right? >> >> >> Looked that way. >> >> >> > 4- A "reallocated sector count" shows the amount of reallocated / >> >> > replaced sectors silently while the drive is operating. When the drive >> >> > has a problem with sector, first it tries to replace that sectors with >> >> > a "spare" sector thus a "reallocated sector" statistic is updated. >> >> >> The reallocation count will increase on any successful >> >> realocation. >> > How much sector will my drive be allowed to reallocate? (eg: What >> > about for a modern 40gb drive ? ) >> >> Depends. It this is really 98 down from 100 and caused by >> 98 reallocation, then a linear extrapolation yields about >> 3000 reallocated sectors, before the SMART status goes bad. >> I expect that at value zero, it will be out of spare sectors, >> but that is pure speculation. With that it could reallocate >> about 4000 sectors. > If a sector is about 512bytes, we can can't we calculate how many they > are? Huh? A sector is either completely bad or readable. No relation to sector size. >> >> Reallocation can happen later. If it was unsuccesful so far, you >> >> get a "pending sector". As son as it has been reallocated, >> >> the pending secor attribut is decreased again. >> >> >> > If the sector is completely unreadable or unreallocatable, you see it >> >> > as "bad" marked in surface scan tools like Seatools. Did i understand >> >> > correct? >> >> >> Yes. >> > That's why the full / surface scan utilities are there. They report >> > "bad" , really "bad" sectors which can't be allocated by firmware of >> > HDD. Right? >> >> They report bad sectors that were not reallocated yet. >> The disk can allways reallocate sectors, but sometimes only >> with dataloss or after several tries. > So, what's the difference between reallocation bad sectors and zero- > fill (low-level sector replacement)? Zero-fill may allow the disk to reallocate a sector it did not manage to reallocate before. > At past, i remember 2 bad sectors were zero-filled(replaced) with full > scan tool Seatools. Also i remember, 32-40kb bad blocks were markend > as bad in Scandisk (chkdsh) automatically. But after formatting and > low-leveling disk, "none" bad sectors are shown during a Dos based > tool Seatool surface scan and chkdsk. The from the read-tests, disk did know which sectors were bad. In the process, they likely were all written to and hence the disk could reallocate then, i.e. move good sectors into their place (logically, not physically). >> >> > 5- Though a drive has "reallocated sectors", if it can read every data >> >> > (checking with several disk reading utilites), it doesn't have a >> >> > unallocatable / unfixable bad sectos which are the reasons of real >> >> > data loss. >> >> >> The reallocated sectros are recognized _and_ corrected problems. >> >> These sectors will not cause problems in the future. Other >> >> sectors may go bad, but if a complete surface scan does not >> >> show any, all data is in sectors that are fine or at least look >> >> like it. In rare cases a sector can work with some data, but be >> >> defect with other data. Very rare with today's disks, due to >> >> heavy use of error correcting codes. >> >> >> You still may want to do a complete surface scan regularly, >> >> so that the disk can recognize sectors slowly going bad >> >> in time and can rescue the data in them. I run such a test >> >> every 14 days automatically, but once every 1-2 months should >> >> do fine. You can make that a part of your standard backup >> >> procedure. >> > I usually do it if i record important data. But still "none" bad >> > sectors have been reported by surface scan of Seatools. >> > So, can i say i don't have any bad sectors present at the moment? >> >> No. But if you have bad sectors, they cannot be detected by a >> read only test. And the probability of such sectors is pretty >> small. > Rod said different? Everybody says different things :-( Isn't there > the truth? For that you have to look to religion. However for modern disks the typical situation is that a bad sector will be recognized in a read-only test. It used to be different, were you would, e.g. have a spot on the disk that could only ttake a zero. If the data in that places then has a zero there, it read fine. If there was supposed to be a one, it could be recognized as defect. > Again, rewinding the topic (seems that topic will break group > record :-) ), what do you mean by saying read-only test? Read it, and see whether that works. A read-write test would write different patterns to a secor ans reqd them back. > Why can't i see if there bad sectors? What is full surface scan tool > used for then??? As I said, there likely are no bad sectors left that have not been reallocated. Arno
From: Arno Wagner on 15 Nov 2007 05:15
Previously kimiraikkonen <kimiraikkonen85(a)gmail.com> wrote: > On Nov 15, 2:45 am, Arno Wagner <m...(a)privacy.net> wrote: >> Previously kimiraikkonen <kimiraikkone...(a)gmail.com> wrote: [...] > Rod said different? Everybody says different things :-( Isn't there > the truth? BTW, I (and I suspect many others here) usually do not read what Rod writes. Although from quotes I have seen in postings by others, he at least sometimes gives good advice now. Arno |