From: <ro...@sp...> - 2013-07-11 16:39:08
|
I had a hard drive yesterday I ran smarttools on. It showed one current pending sector count. I ran a short test and it failed on a specific block at 10%. From what I read here, that's not uncommon for a hard drive to stop the test at the first bad block it finds. Looking through the extended error log (smartctl -x /dev/sda) - the regular error log was empty - I found a bunch of errors. All the complete error reports pointed to the same specific block. To verify, I tried hdparm --read-sector <number of sector found above> and it failed. Since I had nothing to loose, I wanted to see if I could change the pending count, so I used hdparm --write-sector <number of sector found above> --yes-i-know-what-i-am-doing /dev/sda. This worked. Now I used hdparm to try to read that block/sector again and I could, although it had no data. Not surprising. I figured I had just forced the firmware of the drive to reallocate a spare block, so the pending count would go to 0, and the reallocated count would be 1. But when I ran smartctl -x /dev/sda again, the reallocated sector count was still 0! The pending count was also 0 as was the Reallocation Event Count and the Offline Uncorrectable. I re-ran the short test, and this time it passed. I even ran badblocks on the whole drive, and no bad blocks were found. Can someone explain what I did/saw? I find it hard to believe I "repaired" the bad sector, but that's the only explanation I can come up with to explain my results. |
From: Greg S. <gr...@sl...> - 2013-07-11 16:47:08
|
Here's how I understand things [which could easily be wrong] - a failed *read* won't get a bad/failing block re-mapped to a "good" block - only failed *writes.* You might have a read that fails only part of the time - so occasionlly it will read properly and all will be well. To force a re-map of the bad block, you'll need to try to write to that block. If the write fails, then it will re-map the block to one of the spares. Before writing to the block, you may want to try to read it and/or verify there's nothing there you actually need. HTH -Greg On Thu, Jul 11, 2013 at 9:38 AM, <ro...@sp...> wrote: > I had a hard drive yesterday I ran smarttools on. It showed one current > pending sector count. I ran a short test and it failed on a specific block > at 10%. From what I read here, that's not uncommon for a hard drive to > stop the test at the first bad block it finds. Looking through the > extended error log (smartctl -x /dev/sda) - the regular error log was > empty - I found a bunch of errors. All the complete error reports pointed > to the same specific block. > > To verify, I tried hdparm --read-sector <number of sector found above> and > it failed. Since I had nothing to loose, I wanted to see if I could change > the pending count, so I used hdparm --write-sector <number of sector found > above> --yes-i-know-what-i-am-doing /dev/sda. This worked. > > Now I used hdparm to try to read that block/sector again and I could, > although it had no data. Not surprising. I figured I had just forced the > firmware of the drive to reallocate a spare block, so the pending count > would go to 0, and the reallocated count would be 1. > > But when I ran smartctl -x /dev/sda again, the reallocated sector count > was still 0! The pending count was also 0 as was the Reallocation Event > Count and the Offline Uncorrectable. I re-ran the short test, and this > time it passed. I even ran badblocks on the whole drive, and no bad blocks > were found. > > Can someone explain what I did/saw? I find it hard to believe I "repaired" > the bad sector, but that's the only explanation I can come up with to > explain my results. > > > > > > ------------------------------------------------------------------------------ > See everything from the browser to the database with AppDynamics > Get end-to-end visibility with application monitoring from AppDynamics > Isolate bottlenecks and diagnose root cause in seconds. > Start your free trial of AppDynamics Pro today! > http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk > _______________________________________________ > Smartmontools-support mailing list > Sma...@li... > https://lists.sourceforge.net/lists/listinfo/smartmontools-support > |
From: <ro...@sp...> - 2013-07-11 17:18:08
|
I did the read (hdparm --read-sector), it failed, so I wrote to the block. I agree this should force the reallocation/re-mapping. But according to smart, it didn't. The reallocation count remained at 0. That's the part that doesn't make sense. Shouldn't the reallocation/re-mapping increase the count??? BTW, I just (about 5 minutes ago) had an identical experience with different HD. I unreadable [pending] sector. Wrote to it and the reallocation is still 0. In both cases, these were drives I was planning on reformatting anyway, so I took the opportunity to experiment... > Here's how I understand things [which could easily be wrong] - a failed > *read* won't get a bad/failing block re-mapped to a "good" block - only > failed *writes.* > > You might have a read that fails only part of the time - so occasionlly it > will read properly and all will be well. > > To force a re-map of the bad block, you'll need to try to write to that > block. If the write fails, then it will re-map the block to one of the > spares. > Before writing to the block, you may want to try to read it and/or verify > there's nothing there you actually need. > > HTH > -Greg > > > > On Thu, Jul 11, 2013 at 9:38 AM, <ro...@sp...> wrote: > >> I had a hard drive yesterday I ran smarttools on. It showed one current >> pending sector count. I ran a short test and it failed on a specific >> block >> at 10%. From what I read here, that's not uncommon for a hard drive to >> stop the test at the first bad block it finds. Looking through the >> extended error log (smartctl -x /dev/sda) - the regular error log was >> empty - I found a bunch of errors. All the complete error reports >> pointed >> to the same specific block. >> >> To verify, I tried hdparm --read-sector <number of sector found above> >> and >> it failed. Since I had nothing to loose, I wanted to see if I could >> change >> the pending count, so I used hdparm --write-sector <number of sector >> found >> above> --yes-i-know-what-i-am-doing /dev/sda. This worked. >> >> Now I used hdparm to try to read that block/sector again and I could, >> although it had no data. Not surprising. I figured I had just forced the >> firmware of the drive to reallocate a spare block, so the pending count >> would go to 0, and the reallocated count would be 1. >> >> But when I ran smartctl -x /dev/sda again, the reallocated sector count >> was still 0! The pending count was also 0 as was the Reallocation Event >> Count and the Offline Uncorrectable. I re-ran the short test, and this >> time it passed. I even ran badblocks on the whole drive, and no bad >> blocks >> were found. >> >> Can someone explain what I did/saw? I find it hard to believe I >> "repaired" >> the bad sector, but that's the only explanation I can come up with to >> explain my results. >> >> >> >> >> >> ------------------------------------------------------------------------------ >> See everything from the browser to the database with AppDynamics >> Get end-to-end visibility with application monitoring from AppDynamics >> Isolate bottlenecks and diagnose root cause in seconds. >> Start your free trial of AppDynamics Pro today! >> http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk >> _______________________________________________ >> Smartmontools-support mailing list >> Sma...@li... >> https://lists.sourceforge.net/lists/listinfo/smartmontools-support >> > ------------------------------------------------------------------------------ > See everything from the browser to the database with AppDynamics > Get end-to-end visibility with application monitoring from AppDynamics > Isolate bottlenecks and diagnose root cause in seconds. > Start your free trial of AppDynamics Pro today! > http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________ > Smartmontools-support mailing list > Sma...@li... > https://lists.sourceforge.net/lists/listinfo/smartmontools-support > |
From: Greg S. <gr...@sl...> - 2013-07-11 17:41:46
|
Ah, sorry. Somehow I missed the part where you wrote the block...:-) No answers from me in that case. I'm as eager as you, to know the response. On Jul 11, 2013 10:18 AM, <ro...@sp...> wrote: > I did the read (hdparm --read-sector), it failed, so I wrote to the block. > I agree this should force the reallocation/re-mapping. But according to > smart, it didn't. The reallocation count remained at 0. That's the part > that doesn't make sense. Shouldn't the reallocation/re-mapping increase > the count??? > > BTW, I just (about 5 minutes ago) had an identical experience with > different HD. I unreadable [pending] sector. Wrote to it and the > reallocation is still 0. > > In both cases, these were drives I was planning on reformatting anyway, so > I took the opportunity to experiment... > > > Here's how I understand things [which could easily be wrong] - a failed > > *read* won't get a bad/failing block re-mapped to a "good" block - only > > failed *writes.* > > > > You might have a read that fails only part of the time - so occasionlly > it > > will read properly and all will be well. > > > > To force a re-map of the bad block, you'll need to try to write to that > > block. If the write fails, then it will re-map the block to one of the > > spares. > > Before writing to the block, you may want to try to read it and/or verify > > there's nothing there you actually need. > > > > HTH > > -Greg > > > > > > > > On Thu, Jul 11, 2013 at 9:38 AM, <ro...@sp...> wrote: > > > >> I had a hard drive yesterday I ran smarttools on. It showed one current > >> pending sector count. I ran a short test and it failed on a specific > >> block > >> at 10%. From what I read here, that's not uncommon for a hard drive to > >> stop the test at the first bad block it finds. Looking through the > >> extended error log (smartctl -x /dev/sda) - the regular error log was > >> empty - I found a bunch of errors. All the complete error reports > >> pointed > >> to the same specific block. > >> > >> To verify, I tried hdparm --read-sector <number of sector found above> > >> and > >> it failed. Since I had nothing to loose, I wanted to see if I could > >> change > >> the pending count, so I used hdparm --write-sector <number of sector > >> found > >> above> --yes-i-know-what-i-am-doing /dev/sda. This worked. > >> > >> Now I used hdparm to try to read that block/sector again and I could, > >> although it had no data. Not surprising. I figured I had just forced the > >> firmware of the drive to reallocate a spare block, so the pending count > >> would go to 0, and the reallocated count would be 1. > >> > >> But when I ran smartctl -x /dev/sda again, the reallocated sector count > >> was still 0! The pending count was also 0 as was the Reallocation Event > >> Count and the Offline Uncorrectable. I re-ran the short test, and this > >> time it passed. I even ran badblocks on the whole drive, and no bad > >> blocks > >> were found. > >> > >> Can someone explain what I did/saw? I find it hard to believe I > >> "repaired" > >> the bad sector, but that's the only explanation I can come up with to > >> explain my results. > >> > >> > >> > >> > >> > >> > ------------------------------------------------------------------------------ > >> See everything from the browser to the database with AppDynamics > >> Get end-to-end visibility with application monitoring from AppDynamics > >> Isolate bottlenecks and diagnose root cause in seconds. > >> Start your free trial of AppDynamics Pro today! > >> > http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk > >> _______________________________________________ > >> Smartmontools-support mailing list > >> Sma...@li... > >> https://lists.sourceforge.net/lists/listinfo/smartmontools-support > >> > > > ------------------------------------------------------------------------------ > > See everything from the browser to the database with AppDynamics > > Get end-to-end visibility with application monitoring from AppDynamics > > Isolate bottlenecks and diagnose root cause in seconds. > > Start your free trial of AppDynamics Pro today! > > > http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________ > > Smartmontools-support mailing list > > Sma...@li... > > https://lists.sourceforge.net/lists/listinfo/smartmontools-support > > > > > |
From: Martin B. <Mar...@ic...> - 2013-07-12 12:24:30
|
Hi, > Can someone explain what I did/saw? I find it hard to believe I "repaired" > the bad sector, but that's the only explanation I can come up with to explain my results. On writing, the hdd first tries the currently unreadable sector. If that works, pending count goes down without requiring a sector to be remapped. For older hard disks a typical cause for your experience would be bad data in a sector caused for example by power off right in the middle of a write. You'd see a sector that - while physically OK - would always fail reads. A simple write to such a sector would work and repair the failure. Bye, Martin No Disclaimer.icomedias |