Long Term Hard Drive Storage
in [Storage]
|
Prev: Anyone ever had an SATA cable cause data corruption?
Next: Recovering ownerships and permissions after migration
From: Franc Zabkar on 8 Apr 2008 01:06 On Mon, 07 Apr 2008 11:30:27 GMT, svolaf(a)partitionsupport.com (Svend Olaf Mikkelsen) put finger to keyboard and composed: >On Sun, 06 Apr 2008 07:40:37 +1000, Franc Zabkar ><fzabkar(a)iinternode.on.net> wrote: > >>I recently tried to read an old Quantum 120MB HD that was working fine >>when I put it away in my cupboard many years ago. Now I'm seeing a >>whole bunch of read errors. > >Which BIOS or ATA error codes? I thought it was one of these but I just Scandisked them with a thorough surface scan and found no errors: Model ID QUANTUM LPS210A Parameters 723 cyl, 15 heads, 38 sect/track LBA Sectors 412109 Buffer 98 KB (Dual Ported, Read Ahead) Max. PIO Transfer Mode PIO 3 Max. UDMA Transfer Mode Not Supported Unformatted Capacity 201 MB Model ID QUANTUM ELS127A Parameters 919 cyl, 16 heads, 17 sect/track LBA Sectors 0 Buffer 32 KB (Dual Ported, Read Ahead) Max. PIO Transfer Mode PIO 2 Max. UDMA Transfer Mode Not Supported Unformatted Capacity 122 MB They still have my old DOS 6.22 and Windows 3.1/3.11 stuff on them. The software was installed in 1995 and 1997, but the HDDs looks like they may have been manufactured in 1994 and 1992, respectively (judging by the date codes on the ICs). The 210MB HD has noisy bearings but the other one still runs quiet. I suspect I threw the faulty HDD away, so I can't say exactly how I determined that it had bad sectors. Anyway, I believe it was a Quantum LP drive of some sort, so it would have predated SMART. - Franc Zabkar -- Please remove one 'i' from my address when replying by email.
From: Svend Olaf Mikkelsen on 8 Apr 2008 13:59 On Tue, 08 Apr 2008 15:06:53 +1000, Franc Zabkar <fzabkar(a)iinternode.on.net> wrote: >I thought it was one of these but I just Scandisked them with a >thorough surface scan and found no errors: > > Model ID QUANTUM LPS210A > Parameters 723 cyl, 15 heads, 38 sect/track > LBA Sectors 412109 > Buffer 98 KB (Dual Ported, Read Ahead) > Max. PIO Transfer Mode PIO 3 > Max. UDMA Transfer Mode Not Supported > Unformatted Capacity 201 MB > > Model ID QUANTUM ELS127A > Parameters 919 cyl, 16 heads, 17 sect/track > LBA Sectors 0 > Buffer 32 KB (Dual Ported, Read Ahead) > Max. PIO Transfer Mode PIO 2 > Max. UDMA Transfer Mode Not Supported > Unformatted Capacity 122 MB > >They still have my old DOS 6.22 and Windows 3.1/3.11 stuff on them. >The software was installed in 1995 and 1997, but the HDDs looks like >they may have been manufactured in 1994 and 1992, respectively >(judging by the date codes on the ICs). The 210MB HD has noisy >bearings but the other one still runs quiet. > >I suspect I threw the faulty HDD away, so I can't say exactly how I >determined that it had bad sectors. Anyway, I believe it was a Quantum >LP drive of some sort, so it would have predated SMART. > >- Franc Zabkar One reason I asked for error codes, is that for some old ATA disks, a newer BIOS (or operating system or external disk box) is not able to detect the disk, or does not detect the disk correctly. There are different variants of this problem. One example is a "Conner Peripherals 40MB - CP3046" disk. The system BIOS sees the disk, but reports an error, and does not provide a BIOS disk number. The disk can be accessed using ATA commands. Using the Findbad program, I have for the first cylinders: FindBad, version 1.6. Copyright Svend Olaf Mikkelsen, 2007. Searches for bad sectors. OS: DOS 7.10 Disk: PM Cylinders: 65 Heads: 32 Sectors: 40 MB: 41 IDE CHS: 1053/2/40 CTM: 1053/2/40 IDE MB: 41 User sectors: 84240 Start cylinder: 0 End cylinder: 10 First Only the first bad sector in each cylinder will be shown. --------- CHS ----- LBA Code 0 0 27 26 10 1 0 27 1306 10 2 0 27 2586 10 3 0 27 3866 10 4 0 27 5146 10 5 0 27 6426 10 6 0 27 7706 10 7 0 27 8986 10 8 0 27 10266 10 9 0 27 11546 10 10 0 27 12826 10 This is ATA error code hexadecimal 10 "ID not found". The disk however is OK (as new), and the file system can be read correctly. The disk just needs to be accessed using a 4 heads, 26 sectors translation. The disk probably has a "disk type" number, which could be entered in a BIOS from that time, or the 4 heads, 26 sectors geometry could be manually set. Generally I would say, that if important data are to be read from an old disk, it should be done using read only methods, and not using an operating system that will update file access times during read. This could introduce bad sectors. For the disk in question, the output from my Identify program is as follows. The geometry reported cannot be used for accessing the disk. Identify, version 1.3. Primary Master 0 0x0A5A General configuration bit significant information 0 Bit 0: Reserved 1 Bit 1: Retired 0 Bit 2: Response incomplete 3 Bit 5-3: Retired 1 Bit 6: 1=not removable controller and/or device 0 Bit 7: 1=removable media device 0A Bit 14-8: Retired 0 Bit 15: 0=ATA device 1 1053 Number of logical cylinder 2 0 Specific configuration 3 2 Number of logical heads 4 25297 Retired 5 617 Retired 6 40 Number of logical sectors per logical block 7 50 Reserved for assignment by the CompactFlash Association 8 12 Reserved for assignment by the CompactFlash Association 9 0 Retired 10 Serial number (20 ASCII characters) 00000000000000000000 20 3 Retired 21 64 Retired 22 4 Obsolete. (Vendor specific bytes at read/write long) 23 T2.28B Firmware revision (8 ASCII characters) 27 Model number (40 ASCII characters) Conner Peripherals 40MB - CP3046 47 0x0040 Bit 15-8: 0x80 Bit 7-0: Maximum multiple 48 0 Reserved 49 0x0001 Capatibilites 01 Bit 7-0: Retired 0 Bit 8: Must be 0 0 Bit 9: Must be 0 0 Bit 10: IORDY may be disabled 0 Bit 11: IORDY supported, 0: IORDY may be supported 0 Bit 12: Reserved for Identify Packet Device 0 Bit 13: Standby timer values are supported 0 Bit 15-14: Reserved for Identify Packet Device 50 0x0000 Capabilities 0 Bit 0: 1 for device specific Standby minimum 0000 Bit 13-1: Reserved 0 Bit 14: Shall be set to 1 0 Bit 15: Shall be cleared to 0 51 0x0000 Obsolete. (PIO timing) 52 0x0000 Obsolete. (DMA timing) 53 0x0000 Bit 0: Word 54-58 valid Bit 1: Word 64-70 valid Bit 2: Word 88 valid 54 0 Number of current logical cylinders 55 0 Number of current logical heads 56 0 Number of current logical sectors per track 57 0 Current capacity in sectors 59 0x0000 Bit 8: Valid Bit 7-0: Current multiple 60 0 Total number of user addressable sectors (LBA mode only) 62 0 Obsolete 63 0x0000 Bit 10-0: Multiword DMA 64 0x0000 Bit 7-0: Advanced PIO modes supported 65 0 Minimum Multiword DMA transfer cycle time per word 66 0 Manufacturer's recommended Multiword DMA transfer cycle time 67 0 Minimum PIO transfer cycle time without flow control 68 0 Minimum PIO transfer cycle time with IORDY flow control 69 0 Reserved (for future command overlap and queuing) 70 0 Reserved (for future command overlap and queuing) 80 0x0000 Major version number 81 0 Minor version number 82 0x0000 Command set supported 83 0x0000 Command sets supported 84 0x0000 Command set/feature supported extension 85 0x0000 Command set/feature enabled 86 0x0000 Command set/feature enabled 87 0x0000 Command set/feature default 88 0x0000 Ultra DMA 89 0x0000 Time required for security erase unit completion Bit 0-7 * 2: 0 90 0 Time required for Enhanced security erase completion 91 0 Current advanced power management value 92 0x0000 Master Password Revision Code 93 0x0000 Hardware reset result 93 0x0000 Acoustic management Bit 15-8: Recommended Bit 7-0: Current 127 0x0000 Removeable Media Status Notification feature set support 128 0x0000 Security status The above interpretation is mostly based on ATA/ATAPI-5 documentation. -- Svend Olaf
From: Justin on 8 Apr 2008 15:22 sam wrote: > Justin <Justin(a)NobecauseIhatespam.net> wrote: >> sam wrote: >> >>> Yep, way to go IMO. >>> >>> >> Now why are DVD-RAMS better for the long term? > > The writes are checked better when they are written. > > They arent necessarily better in the sense of the media chemistry being better in the long term. > > Actually DVD-RAM won't work - not enough space. I'm dealing with 20 gig DV and m2t files.
From: sam on 8 Apr 2008 15:37 Justin <Justin(a)NobecauseIhatespam.net> wrote > sam wrote >> Justin <Justin(a)NobecauseIhatespam.net> wrote >>> sam wrote >>> >>>> Yep, way to go IMO. >>>> >>>> >>> Now why are DVD-RAMS better for the long term? >> >> The writes are checked better when they are written. >> >> They arent necessarily better in the sense of the media chemistry >> being better in the long term. >> >> > > > Actually DVD-RAM won't work - not enough space. I'm dealing with 20 > gig DV and m2t files. Just use 4/8 You dont need to keep writing them, just write the new ones to a pair of DVDs etc.
From: Franc Zabkar on 8 Apr 2008 18:41
On Tue, 08 Apr 2008 17:59:21 GMT, svolaf(a)partitionsupport.com (Svend Olaf Mikkelsen) put finger to keyboard and composed: >On Tue, 08 Apr 2008 15:06:53 +1000, Franc Zabkar ><fzabkar(a)iinternode.on.net> wrote: > >>I thought it was one of these but I just Scandisked them with a >>thorough surface scan and found no errors: >> >> Model ID QUANTUM LPS210A >> Parameters 723 cyl, 15 heads, 38 sect/track >> Model ID QUANTUM ELS127A >> Parameters 919 cyl, 16 heads, 17 sect/track >>- Franc Zabkar > >One reason I asked for error codes, is that for some old ATA disks, a >newer BIOS (or operating system or external disk box) is not able to >detect the disk, or does not detect the disk correctly. There are >different variants of this problem. > >One example is a "Conner Peripherals 40MB - CP3046" disk. The system >BIOS sees the disk, but reports an error, and does not provide a BIOS >disk number. The disk can be accessed using ATA commands. Using the >Findbad program, I have for the first cylinders: > > >FindBad, version 1.6. Copyright Svend Olaf Mikkelsen, 2007. > >Searches for bad sectors. > >OS: DOS 7.10 > >Disk: PM Cylinders: 65 Heads: 32 Sectors: 40 MB: 41 >IDE CHS: 1053/2/40 CTM: 1053/2/40 IDE MB: 41 >User sectors: 84240 > >Start cylinder: 0 End cylinder: 10 First > >Only the first bad sector in each cylinder will be shown. > >--------- CHS ----- LBA Code > 0 0 27 26 10 > 1 0 27 1306 10 .... > 10 0 27 12826 10 > > >This is ATA error code hexadecimal 10 "ID not found". > >The disk however is OK (as new), and the file system can be read >correctly. The disk just needs to be accessed using a 4 heads, 26 >sectors translation. The disk probably has a "disk type" number, which >could be entered in a BIOS from that time, or the 4 heads, 26 sectors >geometry could be manually set. So the translation appears to be ... 2 x 40 x 1053 physical = 4 x 26 x 810 logical = 84240 user sectors If I understand you correctly, the problem is that the Conner drive reports its physical geometry via the IDE-Identify command but uses its logical (sector translated) geometry, whereas my Quantum drives both report and use their logical geometries. As you say, in your case the BIOS needs to be told what the drive's translated CHS values are. I have a listing of the drive table in an old Award BIOS. It does not include the above CHS combination, nor does it have "user defined drive types". I remember hacking these BIOSes to add such unsupported drive geometries. BTW, I tried to locate the CP3046 drive in TheRef, but it isn't listed: http://marina.mfarris.com/theref/files/tref45a.zip >Generally I would say, that if important data are to be read from an >old disk, it should be done using read only methods, and not using an >operating system that will update file access times during read. This >could introduce bad sectors. Sorry, I don't understand what you are saying here. >For the disk in question, the output from my Identify program is as >follows. The geometry reported cannot be used for accessing the disk. > > >Identify, version 1.3. > >Primary Master > > 1 1053 Number of logical cylinder > 3 2 Number of logical heads > 4 25297 Retired > 5 617 Retired > 6 40 Number of logical sectors per logical block > 7 50 Reserved for assignment by the CompactFlash Association > 8 12 Reserved for assignment by the CompactFlash Association > 9 0 Retired > 10 Serial number (20 ASCII characters) > 00000000000000000000 > 20 3 Retired > 21 64 Retired > 22 4 Obsolete. (Vendor specific bytes at read/write long) > 23 T2.28B Firmware revision (8 ASCII characters) > 27 Model number (40 ASCII characters) > Conner Peripherals 40MB - CP3046 > 47 0x0040 Bit 15-8: 0x80 Bit 7-0: Maximum multiple <snip> >The above interpretation is mostly based >on ATA/ATAPI-5 documentation. It seems to me that Conner Peripherals may not have correctly implemented the ATA-1 (?) standard. The spec clearly calls for the drive to report its "default translation mode", not its physical geometry. http://www.t13.org/Documents/UploadedDocuments/project/d0791r4c-ATA-1.pdf =================================================================== 9.9.1 Word 1: Number of cylinders The number of user-addressable cylinders in the default translation mode. 9.9.2 Word 3: Number of heads The number of user-addressable heads in the default translation mode. 9.9.3 Word 4: Number of unformatted bytes per track The number of unformatted bytes per translated track in the default translation mode. 9.9.4 Word 5: Number of unformatted bytes per sector The number of unformatted bytes per sector in the default translation mode. 9.9.5 Word 6: Number of sectors per track The number of user-addressable sectors per track in the default translation mode. =================================================================== - Franc Zabkar -- Please remove one 'i' from my address when replying by email. |