[clug] 3TB (4k sector) disk question
rodneyp at pcug.org.au
rodneyp at pcug.org.au
Fri Sep 7 17:53:50 MDT 2012
I could not see any obvious technology change-over point for utilising
capacity beyond 2TiB. It might have something to do with version/revision of
ATA spec, which I have not investigated.
Eyal's mainboard, that did work, is reporting SATA 2 speeds only (3 Gb/s).
ISTR that early 3 TB drives were SATA 2 anyway, although I never used one.
Western Digital site has some info re their 2.5 & 3 TB drives. They
initially supplied same cf PCI controller that ensured the full capacity
would be utilised. That practice has now been discontinued, on the basis
that suitable mainboards/add-on cards are readily available.
Looks like a case of "try it and see"
Rod
On Wed, 5 Sep 2012 04:01:18 linux-request at lists.samba.org wrote:
> Message: 5
> Date: Wed, 05 Sep 2012 20:01:07 +1000
> From: Eyal Lebedinsky <eyal at eyal.emu.id.au>
> To: linux at lists.samba.org
> Subject: Re: [clug] 3TB (4k sector) disk question
> Message-ID: <504722E3.90609 at eyal.emu.id.au>
> Content-Type: text/plain; charset=UTF-8; format=flowed
>
> It is still not clear to me what the exact failure more was, so for anyone
> interested, here are the details of the controller and the motherboard at
> boot time: ata4: SATA max UDMA/133 abar m2048 at 0xf7286000 port 0xf7286280
> irq 44 scsi8 : ioc0: LSISAS1068E B3, FwRev=011b5600h, Ports=1, MaxQ=277,
> IRQ=16
>
> Messages when the disk was attached to the PCIe controller::
> mptsas: ioc0: attaching sata device: fw_channel 0, fw_id 17, phy 6,
> sas_addr 0x433b592f806b384f scsi 8:0:8:0:
> Direct-Access ATA ST3000DM001-9YN1 CC9D PQ: 0 ANSI: 5 sd 8:0:8:0:
> Attached scsi generic sg8 type 0
> sd 8:0:8:0: [sdi] 4294967294 512-byte logical blocks: (2.19 TB/1.99 TiB)
> sd 8:0:8:0: [sdi] Write Protect is off
> sd 8:0:8:0: [sdi] Mode Sense: 73 00 00 08
> sd 8:0:8:0: [sdi] Write cache: enabled, read cache: enabled, doesn't
> support DPO or FUA sdi: sdi1
> sd 8:0:8:0: [sdi] Attached SCSI disk
>
> Messages when the disk was attached to the motherboard directly:
> ata4: exception Emask 0x10 SAct 0x0 SErr 0x4050000 action 0xe frozen
> ata4: irq_stat 0x00000040, connection status changed
> ata4: SError: { PHYRdyChg CommWake DevExch }
> ata4: hard resetting link
> ata4: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
> ata4.00: ATA-8: ST3000DM001-9YN166, CC9D, max UDMA/133
> ata4.00: 5860533168 sectors, multi 0: LBA48 NCQ (depth 31/32), AA
> ata4.00: configured for UDMA/133
> ata4: EH complete
> scsi 3:0:0:0: Direct-Access ATA ST3000DM001-9YN1 CC9D PQ: 0
> ANSI: 5 sd 3:0:0:0: [sdi] 5860533168 512-byte logical blocks: (3.00 TB/2.72
> TiB) sd 3:0:0:0: [sdi] 4096-byte physical blocks
> sd 3:0:0:0: Attached scsi generic sg8 type 0
> sd 3:0:0:0: [sdi] Write Protect is off
> sd 3:0:0:0: [sdi] Mode Sense: 00 3a 00 00
> sd 3:0:0:0: [sdi] Write cache: enabled, read cache: enabled, doesn't
> support DPO or FUA GPT:Primary header thinks Alt. header is not at the end
> of the disk. GPT:4294967293 != 5860533167
> GPT:Alternate GPT header not at the end of the disk.
> GPT:4294967293 != 5860533167
> GPT: Use GNU Parted to correct GPT errors.
> sdi: sdi1
> sd 3:0:0:0: [sdi] Attached SCSI disk
> The above GPT errors are due to the bad table built earlier. They were gone
> once a new GPT was built.
>
> cheers
> Eyal
>
> On 09/04/12 20:52, Eyal Lebedinsky wrote:
> > Thanks Rod,
> >
> > As you suggest at the end, and as I reported early today, the problem was
> > the SATA controller which did not handle the required protocol (LBA48?).
> >
> > Once connected to the mobo, the disk showed up with the full size.
> >
> > cheers
> >
> > Eyal
> >
> > On 09/04/12 17:51, rodneyp at pcug.org.au wrote:
> >> Hi Eyal,
> >>
> >> I was away yesterday (and not checking my email from the top of Mt
> >> Twynam during a total white-out
> >>
> >> I've never dealt with a drive > 2TB and might not be able to enlighten
> >> you
> >> much further.
> >>
> >> AFAIK the -z switch in gdisk is for, most importantly, erasing sectors 1
> >> to 34 prior to replacing GPT disk-label with an MBR/DOS disk-label. You
> >> need the -n switch to create a new disk GPT label. I suggest you try
> >> that first.
> >>
> >> By way of comparision, my 2TB WD20 EARX reports to hdparm -I
> >>
> >> CHS current addressable sectors: 16514064
> >>
> >> LBA user addressable sectors: 268435455
> >> LBA48 user addressable sectors: 3907029168
> >> Logical Sector size: 512 bytes
> >> Physical Sector size: 4096 bytes
> >> Logical Sector-0 offset: 0 bytes
> >> device size with M = 1024*1024: 1907729 MBytes
> >> device size with M = 1000*1000: 2000398 MBytes (2000 GB)
> >>
> >> and hdparm -V
> >> hdparm v9.37
> >>
> >> Results are fairly analagous, but HDD from different manufacturers
> >>
> >> gdisk 0.8.5 reports
> >>
> >> Logical sector size: 512 bytes
> >> Partition table holds up to 128 entries
> >> First usable sector is 34, last usable sector is 3907029134
> >> Partitions will be aligned on 2048-sector boundaries
> >> Number Start (sector) End (sector) Size Code Name
> >>
> >> 1 2048 22527 10.0 MiB 0700
> >> 5 22528 4216831 2.0 GiB 8200 Linux swap
> >> 10 4237312 37791743 16.0 GiB 8300 Linux/Windows
> >>data 11 37791744 71346175 16.0 GiB FD00 Linux/Windows
> >>data>>
> >> <snip>
> >>
> >> (gdisk continues speaking 512 B when using 4k logical sectors, which I
> >> find a more than a little confusing)
> >>
> >> I see no switch in gdisk for forcing 4k hardware sectors - that would be
> >> more of a hdparm function and fairly risky. AFAIK, the translation to
> >> logical 512 B sectors is "hardwired" in the current crop of Advanced
> >> Format drives. It might take little processing power and have little
> >> performance "impact" The "impact" comes if a file block has to span 2
> >> physical 4 k sectors and partitioning via gdisk s avoids that.
> >>
> >> Your starting point of a USB drive might be the problem. It's possible
> >> there is some jiggery-pokery in the USB-SATA controller that you have
> >> set aside and which did disable the translation at each start-up. It's
> >> possible that OEM have access to some utility for that purpose. You
> >> could try going to the Seagate web site and looking for their
> >> utiility to see whether it provides for the translation being
> >> disabled. There were reports last year that the then current WD USB
> >> drives could not be repartitioned - even from Windows. Yet, as my drive
> >> ilustrates, it is not an issue with a bare SATA drive.
> >>
> >> OTOH, gdisk does default to 4k partition sectors and that ought to give
> >> you
> >> the performance advantage of Advanced Format by aligning to 4k physical
> >> sectors anyway. Your remaining problem then is capacity.
> >>
> >> Another possibilty is limitations in older SATA controllers/mainboard
> >> firmware that might not cope with > 2.1 TB drives - perhaps try a
> >> different mainboard. (I' m using Proliant N40L. Spec suggests it
> >> handles a max of 2 TB / drive, but it is SATA II not SATA III - Hmm)
> >>
> >> ISTR Rainer purchasing a 3 TB WD for his N40L recently. Perhaps he can
> >> report his experience with it.
> >>
> >>
> >> Rod
> >>
> >> On Sun, 2 Sep 2012 12:00:03 linux-request at lists.samba.org wrote:
> >>> Message: 5
> >>> Date: Sun, 02 Sep 2012 19:36:16 +1000
> >>> From: Eyal Lebedinsky <eyal at eyal.emu.id.au>
> >>> To: linux at lists.samba.org
> >>> Subject: Re: [clug] 3TB (4k sector) disk question
> >>> Message-ID: <50432890.6000504 at eyal.emu.id.au>
> >>> Content-Type: text/plain; charset=UTF-8; format=flowed
> >>>
> >>> Thanks Michael,
> >>>
> >>> Yes, I tried gdisk, which is easy to use. However, it seems to inherit
> >>> the size limit from the disk. It does not switch to 4k sectors and I do
> >>> not
> >>> see a way of telling it to do so. I also do not see a way of telling it
> >>> to
> >>> use a larger number of small sectors than the disk advertises. Note how
> >>> 'hdparm -I' knows:
> >>>
> >>> Configuration:
> >>> CHS current addressable sectors: 16514064
> >>> LBA user addressable sectors: 268435455
> >>> LBA48 user addressable sectors: 5860533168 <<<<< 3TB of 512B
> >>>
> >>> sectors
> >>>
> >>> Nevertheless gdisk uses a lower limit of 4294967294 (2TB), same as
> >>> fdisk.
> >>> What gives? Maybe Rod understands (and can explain) where the
> >>> restriction
> >>> comes from.
> >>>
> >>> cheers
> >>>
> >>> Eyal
> >>>
> >>> On 09/02/12 19:17, Michael James wrote:
> >>>> On 01/09/2012, at 6:49 PM, Eyal Lebedinsky wrote:
> >>>>> Today I bought a 3TB disk, which originally was in a USB3 case (now
> >>>>> out
> >>>>> of it).
> >>>>>
> >>>>> When introduced linux (as bare SATA) it seems that a sector size of
> >>>>> 512B
> >>>>> is used, and the disk size is limited.
> >>>>>
> >>>>> 'hdparm -I' knows that this is a native 4k disk:
> >>>>> Model Number: ST3000DM001-9YN166
> >>>>> Logical Sector size: 512 bytes
> >>>>> Physical Sector size: 4096 bytes
> >>>>> Logical Sector-0 offset: 0 bytes
> >>>>> device size with M = 1024*1024: 2861588 MBytes
> >>>>> device size with M = 1000*1000: 3000592 MBytes (3000 GB)
> >>>>>
> >>>>> However parted thinks otherwise (I created one large GPT partition):
> >>>>>
> >>>>> # parted /dev/sdi
> >>>>> GNU Parted 3.0
> >>>>> Using /dev/sdi
> >>>>> Welcome to GNU Parted! Type 'help' to view a list of commands.
> >>>>> (parted) print
> >>>>> Model: ATA ST3000DM001-9YN1 (scsi)
> >>>>> Disk /dev/sdi: 2199GB
> >>>>> Sector size (logical/physical): 512B/512B
> >>>>> Partition Table: gpt
> >>>>>
> >>>>> Number Start End Size File system Name Flags
> >>>>> 1 1049kB 2199GB 2199GB ext2 primary
> >>>>>
> >>>>> Note the line 'Sector size (logical/physical): 512B/512B'.
> >>>>>
> >>>>> How do I make linux use 4k sector size? Or is there a jumper
> >>>>> on the disk? Or a special SATA command to configure the disk?
> >>>>>
> >>>>> Also, should GPT not be able to create very large disks even with 512B
> >>>>> sectors, using the 64bit internal pointers/sizes/etc.? Where is the
> >>>>> disk size limit coming from?
> >>>>
> >>>> Dear Eyal,
> >>>>
> >>>> If you knew enough to partition it GPT you probably know this
> >>>>
> >>>> but I hope it's worth putting put for the list.
> >>>>
> >>>> The 2 TB limit is inherent in Master Boot Record (MBR/MSDOS)
> >>>> partitioning.
> >>>> The field that delimits a partition doesn't have enough bits to
> >>>> describe
> >>>> anything bigger.
> >>>>
> >>>> Either it's not done in terms of logical blocks or you can't change
> >>>> that
> >>>> anyway.
> >>>>
> >>>> Modern disks actually work in 4 KB blocks, but present logically as 512
> >>>> Byte blocks. If you partition a disk NOT on 4 KB boundaries; expect 10
> >>>> x
> >>>> less performance.
> >>>>
> >>>> This is explained quite well here: http://www.rodsbooks.com/gdisk/
> >>>>
> >>>> BTW Rod's gdisk partitioning tool is quite a groovy little utility.
> >>>> Acts like fdisk, writes GPT partition tables.
> >>>>
> >>>> FWIW,
> >>>> michaelj
> >>>
> >>> --
> >>> Eyal Lebedinsky (eyal at eyal.emu.id.au)
--
test
More information about the linux
mailing list