Jorge, On Wed, 17 Aug 2005, Jorge Izquierdo (UAM) wrote: > Thank you Connie, I've already read that but I thought it won't be > necessary to update to 4.1 to achieve the support for filesystems larger > than 2TB and I hoped that somebody on the list gave me some hint to > resolve this issue without the need for update. I do not think you understand. I think if you use the GPT option and the mke2fs -b 4096 when you make the filesystem and the latest errata kernel and not LVM2 then you should be ok. Do you think the above does not work? -Connie Sieh > > But I will try with SL4.1 (or maybe before eith XFS) if there is no > other solution. > > Thanks again > > Jorge > > > > On Wed, 2005-08-17 at 17:30, Connie Sieh wrote: > > Jorge, > > > > On Wed, 17 Aug 2005, Jorge Izquierdo (UAM) wrote: > > > > > Hi everybody, I'm new in the list so I apologize if the mail is not so > > > clear as it should be. > > > > > > I'm having some troubles with SL4.0 trying to configure a filesystems of > > > 5TB in my SCSI storage (Promise vtrack 15110). > > > When the system starts, the output messages shows that my SCSI device is > > > correctly detected and the information about the size of the RAID > > > storage created is right: > > > > > > scsi0 : Adaptec AIC79XX PCI-X SCSI HBA DRIVER, Rev 1.3.11 > > > <Adaptec 39320A Ultra320 SCSI adapter> > > > aic7902: Ultra320 Wide Channel A, SCSI Id=7, PCI-X 67-100Mhz, > > > 512 SCBs > > > > > > (scsi0:A:0): 160.000MB/s transfers (80.000MHz DT, 16bit) > > > Vendor: Promise Model: 14 Disk RAID5 Rev: V0.0 > > > Type: Direct-Access ANSI SCSI revision: 04 > > > scsi0:A:0:0: Tagged Queuing enabled. Depth 4 > > > SCSI device sda: 2532706176 2048-byte hdwr sectors (5186982 MB) > > > SCSI device sda: drive cache: write through > > > sda: sda1 sda2 > > > Attached scsi disk sda at scsi0, channel 0, id 0, lun 0 > > > > > > (as you can see, I'm using the Adaptec 39320A HBA with the driver > > > aic79xx included with SL) > > > > > > But the problems come when I try to configure my storage with fdisk and > > > then I try to create the ext3 filesystem with mke2fs. First of all, > > > fdisk doesn't allow me to create a unique partition with the full > > > capacity of my disk array, but when I select a partition size larger > > > than 4TB it doesn't work properly. Is this normal? It seems the 64 bits > > > support is not reflected at the time of partition creation. > > > > > > So I decided to create 2 partitions (sda1 and sda2 as reflected in the > > > dmesg text) one of 4 TB and the other one with the rest. And here comes > > > the second problem, I create a ext3 FS on the sda partition with: mke2fs > > > -j -m 0 -b 4096 /dev/sda1 > > > And when I mount my new filesystem and I test the mounted filesystems I > > > get the next information: > > > $# mount /dev/sda1 /mnt > > > $# df -h /mnt > > > Filesystem Size Used Avail Use% Mounted on > > > /dev/sda1 2.0T 103M 2.0T 1% /mnt > > > > > > So the size of my partition is only 2TB when I created a 4TB partition > > > with fdisk. Any ideas about this behaviour? Should I create my > > > filesystem in any other way to support larger than 2TB sizes? Is the 64 > > > bits support broken anywhere with SL 4.0? Am I missing something? > > > > >From the Upstream Vendor Update 1 release notes. I realize that you are > > on 4.0 and not 4.1 but the info here is of importance. Note that if you > > have the latest errata kernel you have the kernel from 4.1 . > > ---------------------------------------------------------------------------- > > > > o Scientific Linux 4.1 provides support for disk devices > > that are larger than 2 terabytes (TB). Although there is limited > > support for this feature in the Scientific Linux 4.0 release, > > 4.1 contains many improvements (both in user space programs and > > in the kernel). In general, 4.1 is considered a requirement for > > support of disk devices larger than 2 TB. > > > > Please note the following guidelines and restrictions related to large > > device support: > > > > . Typical disk devices are addressed in units of 512 byte blocks. The > > size of the address in the SCSI command determines the maximum device > > size. The SCSI command set includes commands that have 16-bit block > > addresses (device size is limited to 2 GB), 32-bit block addresses > > (limited to addressing 2 TB), and 64-bit block addresses. The SCSI > > subsystem in the 2.6 kernel has support for commands with 64-bit block > > addresses. To support disks larger then 2 TB, the Host Bus Adapter > > (HBA), the HBA driver, and the storage device must also support 64-bit > > block address. We have tested the QLogic qla2300 driver and the > > Emulex lpfc driver, included in Scientific Linux 4.1, > > on an 8 TB logical unit on a Winchester Systems FX400 (rev. 3.42B and > > above is required). > > > > . The commonly-used MS-DOS partition table format can not be used on > > devices larger than 2 TB. For devices larger than 2 TB, the GPT > > partition table format must be used. The parted utility must be used > > for the creation and management of GPT partitions. To create a GPT > > partition, use the parted command mklabel gpt. > > > > Scientific Linux requires that all block devices be initialized with a valid > > partition table, even if there is a single partition encompassing the > > entire device. This requirement exists to prevent potential problems > > caused by erroneous or unintended partition tables on the device. > > > > . The / and /boot directories must be located on devices that are 2 TB > > in size or less. > > > > . Various issues with LVM2 on large devices are fixed in > > Scientific Linux 4.1. Do not use LVM2 on devices larger than 2 > > TB prior to installing 4.1. > > > > As noted above, Scientific Linux requires that a partition table be written to > > the block device, even when it is used as part of an LVM2 Volume > > Group. In this case, you may create a single partition that spans the > > entire device. Then, be sure to specify the full partition name (for > > example, /dev/sda1, not /dev/sda), when you use the pvcreate and > > vgcreate commands. > > > > . The maximum size disk that can be a member of an md software RAID > > set is 2 TB. The md RAID device itself can be larger than 2 TB. > > Devices have been tested up to 8 TB. > > > > . Various issues with e2fsprogs that occur on devices larger than 4 TB > > are addressed in Scientific Linux 4.1. Prior to 4.1, > > these issues can be worked around by specifying mke2fs -b 4096 when > > making an ext2 or ext3 filesystem. The workaround is not necessary in > > 4.1. > > > > The ext2 and ext3 filesystems have an internal limit of 8 TB. Devices > > up to this limit have been tested. > > > > You may want to use the mke2fs -T largefile4 command to speed up the > > creation of large filesystems. > > > > . The GFS filesystem is limited to 16 TB on 32-bit systems, and 8 > > exabytes (EB) with 64-bit systems. GFS filesystem > > sizes up to 8 TB have been tested. > > > > . NFS partitions greater than 2 TB have been tested and are supported. > > > > . Scientific Linux 4.1 user space tools are compiled > > for large file support. However, it is not possible to test every > > program in this mode. Please file a problem report if issues arise > > when using the tools for large file support. > > > > . The inn program does not function correctly with devices larger than > > 2 TB. This will be addressed in a future release of Scientific Linux. > > > > ---------------------------------------------------------------------------- > > > > So make sure you use mke2fs -b 4096 when making a file system greater than > > 4GB on 4.0 . I think the problem has to do with "wrap around", which > > surely does not seem good. > > > > -Connie Sieh > > > > > > > > Thanks for any help or any suggestion > > > > > > Jorge > > > > > > > > > >