|
Other BSD and UNIX/UNIX-like Any other flavour of BSD or UNIX that does not have a section of its own. |
|
Thread Tools | Display Modes |
|
|
||||
Quote:
AFAIK, you cannot install a *BSD on en ext*** file system. But you can keep the ffs*(OBSD), ufs**(FBSD), zfs**(Solaris) for the ON (Solaris acronyms for Operating system and Networking). I have such a setup with all data, shared files, home, DB, www, distfiles, /usr/ports, /var/db/portsnap, ... on ext2fs systems ("extended partitions"). AFAIK, only need ON to be on the native filesystem. All the rest are data (the stuff which is personal to you and you surely not want to lose). Data can be housed wherever you like, even a thumbdrive. Thumbdrive readable by any ON, just have to chown UID/GID.
__________________
da more I know I know I know nuttin' |
|
||||
Quote:
Quote:
|
|
||||
I suggest you read this section and this wikipedia page.
To put it simple: ad0 -> First PATA Hard Drive using FreeBSDs 'ad' driver. ad0s1 -> first primary partition on ad0 -- also known as a 'slice' ad0s1a -> partition 'a' within the disklabel for ad0s1 and so on. BSD partitioning scheme is different from Linux or DOS and is fully documented. The slice/partition notation should probably be reversed but I didn't invent the damn thing :\. One BSD slice is a lot like an Extended DOS Partition but you are not limted to haveing 1 of them and depending on the BSD flavor, 8-16partitons in side them in a manor comparable to logical drives. Same concept from user point of view. Unless you want to f*** around with things: /boot, /etc, and possibly /bin must be on the / because /boot is needed for bootstrapping. There is probably no sense in moving /sbin off the / either, the space gains not worth the bother unless your back in 1984. Typical convention is to create either an a and b partition (/ and swap within the slice) or to create several partitions within one or more slices. this is my OpenBSD systems layout: Code:
Terry@vectra-$ df -hl Filesystem Size Used Avail Capacity Mounted on /dev/wd0a 147M 30.4M 110M 22% / /dev/wd0h 393M 938K 372M 0% /home /dev/wd0d 98.3M 6.0K 93.4M 0% /tmp /dev/wd0g 6.7G 400M 6.0G 6% /usr /dev/wd0e 148M 10.6M 130M 8% /var /dev/wd1a 11.8G 96.0M 11.1G 1% /usr/local /dev/wd1d 44.3G 12.2G 29.9G 29% /srv Terry@vectra-$ (at least modern) BSD systems generally name devices as driver major number [additional information]. i.e. xl0 -> first network interface using 'xl' driver, there is no eth0, eth1, etc -> it's all named by driver. ad0 -> first pata drive on FreeBSD da0 -> first scsi/usb drive on FreeBSD wd0 -> first pata drive on OpenBSD sd0 -> first sci/usb drive on OpenBSD etc. FreeBSDs file system uses soft updates and is gaining journaling support -> soft updates is a different way of filling the same need that journaling does. Using ext3 under FreeBSD is mounted as ext2 -> and that means mounted without journaling so there is no point in using ext2/ext3 as a FreeBSD system slice. UFS is better suited for system data. Use FAT32/EXT2 for data to be shared with other OSes read+write, as ReiserFSv3 and JFS == read only in FreeBSD.
__________________
My Journal Thou shalt check the array bounds of all strings (indeed, all arrays), for surely where thou typest ``foo'' someone someday shall type ``supercalifragilisticexpialidocious''. Last edited by TerryP; 17th May 2008 at 06:34 PM. |
|
||||
Quote:
UNIX came out long before DOS. UNIX used slices (of the disk) and partitions (in the slices) long before the PC came out. Microsoft and IBM butchered things when they released DOS on the first PC. When BSD was ported to the i386, they brought along the UNIX heritage. |
|
||||
True but unfortantly (most of us/) we are stuck on PC's.
Which for me, should we say the PC has quite a bit of a legacy DOS feel to some parts of it's design and implementation when I think about it ;-) I just prefer how BSD Unix handles the situation, the notation is learnable hehe.
__________________
My Journal Thou shalt check the array bounds of all strings (indeed, all arrays), for surely where thou typest ``foo'' someone someday shall type ``supercalifragilisticexpialidocious''. |
|
||||
A slice is not exactly a parition.
On Wintel, you have 4 partition, or 3 primary partitions and one "extended" partition chaining logical partitions one to the preceeding one. A slice in *BSD (and *BSDs habdle this differently) are a "primary" partitions holding several "logical" partitions (called labels). In one BSD (primary) partition, you house 16 labels. More exactly, 8 OS labels, and 8 "free" hooks to alien labels. I.e.: the first "alien" label, usually a Windows FAT or NTFS label will be recognized as label "i", the "j", ... FreeBSD 8 now recognizes 26 labels (as the 26 letters of the alphabet (only limitation), so it can handle non-ufs2/zfs file systems (aka "logical" partitions) from letter "i" to "z" = 18. So (as "b" is reserved to the swap. "c" descibes the partition volume, "a" is resrved to / root), you can handle 3 partitions with 6 "logical" partitions, plus 18 "logical" partitions. Reason why these are called slices: 3 slices (holding 6 labels) plus 18 labels. Or 2 slices (each with 6 labels) and 2 slices each with 18 labels. Whereas, Wintel only handles "primary" (one partition, one label) and an "extended" partition holding an "unlimited?" number or chaines "logical" partitions. Clear now? Clear as mud.
__________________
da more I know I know I know nuttin' |
|
||||
well, i know that in linux, there can be a max of 4 primary partitions...
(i believe that this is a standard limitation of x86 machines...) now, of those 4 primaries, you can make a logical partition the logical partition is a primary partition, but is read so as that that partitions WITHIN the logical partition, are unlimited so, a logical partition can hold several partitions, thus breaking the 4 primary partition limit originally created... |
|
||||
Quote:
Pawel is working on ZFS version 10 support in his Perforce branch. Once he completes writing the regressions tests, he will commit it to -CURRENT. Depending on the timeline, it may make it into FreeBSD 8.0. Info on his talk from BSDCan 2008: http://kerneltrap.org/FreeBSD/BSDCan_2008_ZFS_Internals ZFSv10 allows you to use ZFS without a separate / (root) partition. MacOS X 10.5 includes ZFS support as well. AFAIK, no other OSes support ZFS. |
|
|||
What is ZFS and why is it better than FFS/UFS?
__________________
And the WORD was made flesh, and dwelt among us. (John 1:14) |
|
|||
While most of us are willing to answer questions, it would be helpful if you did some initial research yourself.
I'm not an expert on the subject, but, here are some articles I found via Google. http://en.wikipedia.org/wiki/ZFS http://www.opensolaris.org/os/community/zfs/whatis/ http://www.opensolaris.org/os/community/zfs/ Also, the FreeBSD specific Wiki: http://wiki.freebsd.org/ZFS Note: Better is subjective, personally, ZFS will never replace my UFS/FFS partitions. |
|
||||
For end users that do not have demanding storage requirements and/or disk read/write bottlenecks related to iowait, ZFS means nothing and is no better than FFS/UFS(or any other file system for that matter). ZFS is Sun's forward looking "Future Proof" filesystem for SysAdmins that need performance, data integrity and "Unlimited scalability". I can certainly see the benefits of ZFS for the Enterprise(Sun's Primary Market) but again, for the end user running FreeBSD on his/her desktop(or even a personal server) its not important at this stage.
|
Tags |
zfs |
Thread Tools | |
Display Modes | |
|
|