partitioning a 60GB laptop

i have a laptop with a 60GB hard drive. i am installing openBSD on it and there will be no other OS's present on this machine (i.e. the entire drive is to be used by openBSD). I am an average user and by that i mean i listen to music (have a lot of music), watch dvds, use Office programs, browse the web, check email, experiment... would it be beneficial for me to create more than one partition? if so what do you reccomend?

See this URL:
http://www.openbsd.org/faq/faq4.html#SpaceNeeded
As a general rule of thumb, this is what I do:

Of course, these are rule of thumb and I vary them according to my needs. For instance, for a webserver, I would set /usr to say 4 GB and make /var/www a separate partition with the rest of the space allocated to it. For a DNS/mail server, I use djbware and tend to allocate a separate /var/djb partition for DJB software alone. In your case, you may want to keep /usr to something like 10 GB and allocate the rest to /home, or perhaps a separate /mp3 partition for just your mp3s.

i'm still installing openBSD 3.6... i have partitioned my hard drive using a similar scheme... thanks... i am now using the ftp method of installation... will the downloaded files be installed in the appropriate partitions? (i.e. does the installer know where to place these files and to look for other mounted partitions or did i overlook an opportunity to specify?)

i really not agree.
you have 60GB, why so avaricious with partitions?
do you like probls?

- long time i worked w/ bsd, but you should be able to
put / /tmp /var an /usr on the same partition, make it
large enougt, say >4GB

- remember they are all SYSTEM directories, you (as normal
user) never should be forced to write on them.

- don't be stingy w/ swap, may be later you install more RAM
so the formula: 2x available physical RAM will be wrong!

- distribute the rest on partition w/ size you can
efficiently backup, assumed the backup device works
on max 10GB and you have a 50GB to save ..... no problem,
with a little (more) unix knowledge

- NEVER install foreign progs and data in /usr, because
a) offen you need root perms to work on
b) in the stupid case, you must re-install, /usr and
all your foreign progs are GONE.

- install foreign progs and data in an other partition,
classic: /export, if you need, make a symlink from /usr
to that dir (cd /usr; ln -s ../export/xxx), in case of
re-install, you sure can avoid to format += newfs the part
mounted on /export, this way you SAVE the DATA

- on OLD bsd, /usr have to mandatory be on partition 'g'
maybe it's no longer the case.

It is a very bad idea to put / along with /tmp, /var and /usr. / contains /bin and /sbin. The ideal would be that once the OS boots up, it should have to access the / partition very sparingly. Thus, if you have a disk crash, chances are that the drive head will be positioned over a different area on the disk and thus your / partition will be intact. Therefore, you can in theory use the tools in /bin and /sbin (which are in the / partition) to bring up the rest of the system.


/var is usually read to and written to a lot, especially if this is a mail server. /tmp is written to by a lot of system daemons also. If I put them as separate partitions, I can mount / as read-only for greater security .


Notice I had a cap in my swap size. Actually, 512 MB may be too much too , because it never gets used up in most of my configurations. Heck, I could probably get away with just 256 MB for my purposes. Of course, your mileage may vary. Kernel core dumps do get written to the swap area, so if you're interested in debugging it, make sure that swap is at least slightly greater than available RAM.


This is very good advice.


OpenBSD standard is to install foreign programs under the /usr/local tree. This is pretty much the standard in all BSDs and Linux as well (man 7 hier or http://www.pathname.com/fhs/, though some Linux distros and howtos don't follow this at all. Neither does Daniel J. Bernstein). NetBSD standard is to install source compiled stuff in /usr/local and anything installed from ports/packages under /usr/pkg tree.


Not any more: http://www.openbsd.org/faq/faq4.html#Disks

Yes, the installer knows what partitions to install stuff at.

how should i edit my fstab? when reading through the man pages i noticed an example that included lines for /proc and /kern both of which i have not included in my fstab. do i need to include these? the example also had a separte partition for /etc and /etc/local. do you see this as beneficial?

what does it mean to dump a filesystem? does this mean that data will be erased?

scorpion, still not agree, we sure are NOT speaking about the same unix.
i work on solaris, and never backup / /usr /var /tmp because they are on CD.
to restore them i need 20 mins.
my private data, i mean /home and foreign progs are on an other part.
i know standard is /usr/local, on my sys /usr/local is a symlink to
/whateveryouwant/local where is the probl?
/tmp is 777
all progs (run by an mortal user) writing in / /usr and /var are 'setuid'
...
:q!


PS: i know OS they will 'swap' on partition '0' (also called 'a')

Haven't worked with Solaris for years, but on *BSD, a lot of stuff is written to /var (such as /var/db, /var/mail/ etc.)

Also, there's a good reason to keep / and /var on separate partitions. We had boxes installed with Linux and RedHat's default partitioning scheme puts everything on one large partition (/, /var/, /usr, /tmp etc. are all in one disk partition). Now we had an app that wrote lots of cache files (and a job to clear them that ran once a week). Also, we had other regular system jobs running on the box, but we turned off sendmail, hence the mail updates generated by the system jobs were being cached in the mail queue /var/mqueue. Eventually, all the jobs ended up filling the disk's inode table (even though the disk had space on it, the inodes were all used up and so no new files could be created). First sign of trouble we noticed was when apache started to return errors to us. So we tried to ssh into the box to clear some files. Now, since all the files in /var and /var/cache/tmp were all taking up the inode space and since everything was on a single partition, we couldn't ssh in? Why? Because ssh creates a file in /tmp when you log in and it couldn't do this because the inode table was full! The solution was to drive to the colocation (45 minute drive) and log into the box and empty out stuff. If we had different partitions, the filling up of inode table on /var wouldn't have affected our logging in, because /tmp is located in a root (or its own) partition.