Thanks for the suggestions... original message below.
The problem with just trying to free enough space on / is...
If you remove files that are going to be replaced anyway during the upgrade
they don't help, only the difference between the old and new version.
Simple answer... bite the bullet, backup everything, repartition, and restore.
partition with average suggestion of 128mb root (up to 256 for future
growth), 1gb /usr, 3x+ram for swap, and 256mb+ for /var.
To cut some corners it may be possible to do some juggling, but in the end,
it may as well be done for 4 partitions, instead of one now and others later.
I may opt for dumping everything to an external disk, booting off that,
then repartitioning and restoring back. May prove a bit quicker and easier
than tape and booting off cdrom.
Original message:
>I'm trying to upgrade to 4.0F and it keeps saying I need more space, even
>after removing a lot from / it still huddles around the same amount...
>700K (even if I remove files of several megabytes).
>
>Filesystem 512-blocks Used Available
>Capacity Mounted on
>/dev/rz0a 126814 78594 35538 69% /
>/proc 0 0 0 100%
>/proc
>/dev/rz0g 4116168 2146080 1558470 58%
>/usr
>/dev/rz0b 253662 105818 122476 47%
>/var
>
>
>I'm assuming it's because the / partition is only 64mb... so I need more
>space...
>
>I'm trying to do this as easy as possible... They are not AdvFS
>
>8 partitions:
># size offset fstype [fsize bsize cpg]
> a: 131072 0 4.2BSD 1024 8192 16 # (Cyl. 0 - 92*)
> b: 262144 131072 4.2BSD 1024 8192 16 # (Cyl. 92*- 276*)
> c: 8888924 0 unused 0 0 # (Cyl. 0 - 6242*)
> d: 0 0 unused 0 0 # (Cyl. 0 - -1)
> e: 0 0 unused 0 0 # (Cyl. 0 - -1)
> f: 0 0 unused 0 0 # (Cyl. 0 - -1)
> g: 4247854 393216 4.2BSD 1024 8192 16 # (Cyl. 276*- 3259*)
> h: 4247854 4641070 swap # (Cyl. 3259*- 6242*)
>
>
>any suggestions? I don't mind taking away space from /usr or /swap, as
>2gb each is probably overkill (256mb ram, swap typically low usage)
>
>Thanks
>--Dan
Here's the long description from DEC on repartitioning....
>TITLE: Repartitioning for v4.0
>Copyright (c) Digital Equipment Corporation 1994. All rights reserved.
>
>OP/SYS: DEC OSF/1 for AXP Version 3.0
>
>SOURCE: Digital Equipment Corporation
>
>
>INFORMATION:
>
>Below are the steps necessary to repartition your root disk.
>
>STEP 1:
>
>Create a full system backup using dump or vdump. Backup all partitions
>on the root disk.
>
>STEP 2:
>
>Obtain, through the system console, the location (the unit
>number) of your CDROM, your system disk, and the tape
>drive that contains your system backup tape. Also
>note the system disk's device type. If you need
>assistance on doing this, refer to your system's hardware
>manuals.
>
>The following is an example on a DEC 3000 Series Processor:
>
> >>> show dev
>
>BOOTDEV ADDR DEVTYPE NUMBYTES RM/FX WP DEVNAM REV
>------- ---- ------- -------- ----- -- ------ ---
>DKA0 A/0/0 DISK 426.25MB FX RZ25 0700
>DKA100 A/1/0 DISK 2.10GB FX RZ28 435E
>DKA400 A/4/0 DISK ..... RM WP RRD42 4.3d
>MKA500 A/5/0 TAPE RM TZK10 01B4
>
>Here, DKA100 is the system disk and it's type is a RZ28.
>DKA400 is the location of the CDROM and MKA500 is the tape device.
>
>STEP 2:
>
>Insert the DEC OSF/1 CD into the CDROM and then boot that CD.
>The actual boot command will depend on the location (unit number)
>of the CDROM and your hardware.
>
> >>> b dka400
>
>Choose the 3rd option on the menu: SYSTEM MANAGEMENT
>For v4.0 and newer, select the "Unix Shell" button.
>
>
>
>STEP 3:
>
>Make the special device files for the system disk and the tape drive.
>You need to use the unit numbers you saw from the console command
>executed in Step 1.
>
># cd /dev
># MAKEDEV rz1 tz5 (tz# - the number will be the unit no of the tape
> device
> rz# - the number will be the unit no of your
> system disk. NOTE: If your system disk
> is on a RAID device, the format will be:
> re# )
>------------------------------
>Digression #1
>If our boot disk, or tape drive isn't on the first bus, the device
>name is determined this way :
> [dm]kaNNN = bus 0, [dm]kbNNN = bus 1, [dm]kcNNN = bus 2
> device # = ( 8 x bus# ) + unit number
> ie if tape name is mkb300, bus is 1, unit number is 3
> ( 8 x 1 ) + 3 = 11
> device name = tz11
>End Digression #1
>------------------------------
>
>STEP 4: Edit the disklabel for the new partition sizes.
>
>Edit the disklabel on your system disk and make the root domain and
>fileset.
>
># disklabel -Fs /dev/rrz1a unused
># disklabel -Fs /dev/rrz1b unused
># disklabel -Fs /dev/rrz1g unused
># disklabel -e rz1
> (You need to replace rz1 with the device you made
> in Step 3 and you must use replace rz28 with
> your system disk device type found from Step 1.)
>------------------------
>Digression #2
>This will put you in the 'ed' editor. You should either be familiar with
>the following commands, have the manual handy, or be prepared to try
>several times before getting it right.
>
>The lines in the partition table info that you need to change start at
>about line 20. To print a line and set the current line to edit, type
>the line number and press return.
>Continue trying line numbers until you find the line for the partition
>to modify.
>When you find it, use the 'c' command to change the line. You will only
>need to enter the partition letter, the size, offset, and filesystem
>type. You don't have to retype the '#(Cyl. xxxxxx)' at the end.
>
>
>A sample session is shown here :
>-------- begin session
># disklabel -e /dev/rrz1a
>973
>20
>8 partitions:
>21
># size offset fstype [fsize bsize cpg]
>22
> a: 131088 0 unused # (Cyl. 0 - 164*)
>.c
> a: 262160 0 unused
>.
>w
>q
>write new label? [y]: y
>#
>-------- end session
>
>Then check your disklabel to make sure it worked as you expected. Use
>the 'disklabel -r <device>' command.
>End Digression #2
>--------------------------
>
># newfs /dev/rrz1a (or 'mkfdmn root_domain rz1;mkfset root_domain root'
> for ADVfs)
>
>STEP 5:
>
>Mount the root filesystem, change directory to its location,
>and then issue the restore command to restore your root filesystem.
>
># mount /dev/rz1a /mnt (mount -t advfs root_domain#root /mnt)
># cd /mnt
># vrestore -xvf /dev/rmt0h ( or restore )
>
>-------------------------
>Digression #4
>The vrestore command shown above is more informative if you add the 'v'
>option. There is also the problem that if the root filesystem isn't the
>first dump on the tape you have to skip forward on the tape to the
>right fileset. Vrestore doesn't implement the 's' option that restore
>has to do this. The 'mt' command is used to position the tape to
>the proper fileset. If you need to restore the /usr fileset as well,
>the 'mt' command _is not_ available when you reboot as described later.
>If you want the 'mt' command available when you restore the /usr
>partition, copy it to /sbin after completing the restore of root.
>
>
>It is _very_ important to know the order of the dump files on the tape.
>If you are unsure, use the 'vrestore -tvf' command to list each one.
>When you have seen enough of the listing to identify which fileset it is
>you can cancel the listing with Ctrl-C, then rewind the tape and use
>mt to move forward to the next file and do a listing on it.
>End Digression #4
>-------------------------
>STEP 6:
>
>When the / (root) filesystem restore is complete, change directories to
>root, dismount the ufs root filesystem, and halt the system.
>
># cd /
># umount /mnt
># halt
>
>STEP 7:
>
>Restore /usr and the other partitions on the root disk, if any.
>
># boot -fl s
>
>STEP 8:
>
>Select system shell or system management.
>
>When you get the # prompt, make the root filesystem read writable and
>then newfs the usr partition. You can find out the device and
>partition that your /usr is on by looking in the /etc/fstab file.
>
># mount -u /
># newfs /dev/rrz1g
># mount /dev/rz1g /usr
>
>STEP 9:
>
>Your last step is to now change your directory to /usr and start the
>restore. When the restore is completed, you can continue to boot
>your system up to multi-user mode.
>
># cd /usr
># vrestore -xv (or restore -x)
># init 3
>
>--------------------------
>Digression #5
>If you don't copy 'mt' from the /sbin directory while booted from the
>cd and you need it during step 9, here are your other options.
> 1. Do several 'vrestore -tf /dev/nrmt?h' commands to skip the
> intervening dump files. This can be very slow on large
> filesets.
> 2. Mount the DUnix bootable installation cd and use the mt
> command found in the isl directory.
> 3. Use dd to read N-1 files to /dev/null :
> dd if=/dev/nrmt?h files=2 >/dev/null
> This will position the next read at fileset 3.
>End Digression #5
--------------------------------------------------------------------------
Dan Kirkpatrick dkirk_at_phy.syr.edu
Computer Systems Manager
Department of Physics
Syracuse University, Syracuse, NY
http://www.phy.syr.edu/~dkirk Fax: (315) 443-9103
--------------------------------------------------------------------------
Received on Fri Jan 07 2000 - 14:23:59 NZDT