hi folks,

just a quick question here.
have currently set up my first samba file and print server on the working place last week.
(thanks for everybody on the web and lq in particular who gave me the support to learn how to do this)

i'm currently running this slack on an old box, making sure i still had my working xp server as a back up, but since everything is running smoothly, i want to set up the server on the box that is ment for it.

at this moment it is a simple file and print server with the following partition scheme.
/root
/swap
since there is only 10 GB of space on this old back up server this seemed like the most smart thing to do, but the actual server box which i will install this weekend has plenty of hard drive, so i was thinking about the following partitions.

/boot 100 MB
/swap 1 gig
/root 10 gig
/var 2 gig
/home +- 150 gig

this would be OK i guess for a file and print server, but i might want to turn it eventually also into for instance a database, web and or mail server.

so the question is...... would this partition scheme be OK for such a thing? or would it be smart to save some GB's for future applications some where else?

much obliged!!!

Ouch 150 GB for /home ?

Maybe you could use more partitions for various usages
(/music /videos /backups ...)

With so much disk space, I like to have LOTS of partitions. I have something like:
swap 1GB

/ 3GB
/tmp 1GB
/var 1GB
/usr 5GB
/home 10GB

I repeat the pattern for the five install partitions a couple times so I can easily have several independent installations -- a set ready-to-go for the next Slack version and a set for a test install where I can try things out and mess up without hurting my production install.

Then I have a couple big 20+ GB partitions for archiving old files or music or whatever.

Is the 20-partition limit for what Linux is comfortable with still true? Is that just the limit for splitting up one hard drive?

From man fdisk :

63 partitions for IDE? But does that mean all 63 can be primary? Or just exactly what is the limit to the number of primary and logical/extended partitions are there?

I have been an advocate of disk partitioning going way back to my MS-DOS days. Even back then I had three partitions: C: for system stuff, D: for all of my applications and programs, and E: for all of my data. And that was back in the days when hard drive storage space was measured in MBs, not GBs or TBs.

When I migrated to Windows I used the same basic process and on my current NT4 box I have C:, D:, E:, K:, L:, and O: partitions. I also have a second NT system partition I use as a back door to my primary system partition. For conversation's sake, I still have my original WFWG 3.11 setup installed on a small 512 MB partition.

Three years ago when I began tinkering with GNU/Linux I continued my appreciation for partitions. On my 40GB hard drive, currently I have all of Windows partitions and three distros. I no longer use two of the distros and probably should one day remove them to provide drive space in addition to the space I already don't use. As Geist3 mentioned, I could use that extra space for reckless experimentation. I maintain a separate user account to do something similar, such as installing software (locally) that I am unsure about, or testing configuration changes.

Similar to my Windows setup with an alternate boot partition, I also duplicate my Slackware installation to serve as a simple back door to booting into my system. As related to your question, my current GNU/Linux partitioning (ignoring several partitions):

swap 256 MB
/boot 64 MB
/home 4 GB
/tmp 256 MB
/usr/local 2GB
/opt 2GB
Slackware /var 256 MB
Slackware /usr 2 GB
Slackware / 256 MB
Slackware /var (duplicate) 256 MB
Slackware /usr (duplicate) 2GB
Slackware / (duplicate) 256 MB

Partitioning provides me flexibility. Should I desire to perform a clean install, I need not reinstall everything in /boot, /home, /usr/local, or /opt. Very nice. And back in my early days of experimenting with several distros before deciding to stick with Slack, that decision saved me time and potential grief. Disk partitioning will not prevent heartache in the event of a hardware failure, but provides a nice layer of protection otherwise.

The only limitation I can recall is back when I tried installing Knoppix 3.x to a partition beyond the 15th partition. This is an IDE drive and as keefaz noted from the fdisk man page, I should have been able to install at a higher partition number but no go. Otherwise I have been delighted to find that GNU/Linux distros simply do not care where they are located as long as the boot loader is configured correctly.

I hope this helps.

P.S. Anticipating comments about my duplicate Slackware partitions, yes, my backup scripts ensure that fstab is correct for my duplicate partitions. And I use a separate GRUB boot option if I ever need to get into Slack that way. Since I maintain a separate boot partition, I seldom need tinker with that too.

mmmm... sometimes it's easier having only 10 GB's, at least that limits your options a little!!!!


you're right, i only have 7 GB's off files at the moment, so if i would use +- 50 GB that would be enough for years to come, don;t have any use for music and stuff, since its for an office environment and i use another machine for back ups....
but where does for example MySQL leave its data???


don't get me started on constantly tinkiring my box, it's not broken? lets fix it anyway!!!!.... Butt!! since its a production server i would like to keep it as simple and steady as possible, also with the idea of only being able to maintain it remotely in the future.
keeping a double set up just in case might not be a bad idea though.
would it be OK to leave some space empty so i could add it later on a RAID1 disc????? or do you need to have all final and definitive partitions in place before you create the raid?


thought there was a limt of 4? primary patitions.. note really sure though

yes!!! it did, but the more i know, the more questions i have. don;t know if you can spare a thought on the remaining questions above.




thanks !!!!

You will definitely want to have /tmp on it's own partition. This is because it is world writable, so you want to be able to mount the partition with the "noexec" option. Any world writable partition should use this option. It also makes sense to put /usr/local on its own partition. This is the hierarchy structure base for "local" software, and it is guaranteed that a distro upgrade or a re-installation will not touch this directory.

The /var directory is commonly in it's own partition. This is to prevent a hacker from crashing your server by flooding it, causing logs to grow to the point where there is no spare space left on the partition. If /var/logs is located on the root partition, this can cause it to crash.

Not sure about RAID because I have never used RAID. But I think if you are concerned that in the future your partitions "shrink" to being too small, then that is what Logical Volume Management (LVM) is about (http://tldp.org/HOWTO/LVM-HOWTO/index.html).

As far as low level partitioning I think four primary partitions is still the limit allowed in any operating system. However, one of those primary partitions may be an extended partition that can be subsequently divided into logical partitions. I don't know how the 63/15 limit is counted, but my guess is a simple count---ignoring whether a partition is a primary or extended/logical and just counting total partitions.

Of course on x86 hardware, you couldn't have more than 4 primary partition

The kernel documentation says (Documentation/devices.txt) :
But you have 63 available partitions in a disk
(/dev/hda1.../dev/hda63)

i should have known!!! now i have more questions than i can handle for this evening,
but hey....thats why i like slack!

will just give it a try tomorrow and see how it works out experimenting with adding partions after creating the RAID and resizing these with LVM.
thought i would just simply do another slack install in the weekend, but now it might take a little longer, but i'm not in a rush anyway so who cares.

so after some sleep i will try the following.

/boot 100 MB
/swap 1 gig
/root 10 gig
/usr/local 5 gig ?
/tmp 1 gig
/var 2 gig
/home +- 100 gig
spare +- 50 gig

then install slack and create RAID, play a litte with the LVM to try to see if i can add the spare gig's somewhere and remove them later, just too see how it works.


i'm keeping a log of this installation and will post this when i think i have everything figured out. (this will probably raise a lot more questions again!!!)

thanks all for now!!

here we are again!!

had some good night sleep and a great breakfast, ready to proceed with the plan
1) had my slackware cd1 in place,
2) my floppy with painfully made tagfiles was ready to go,
3) got some sandwiches from the local deli,
4) a fresh pack of brandaris for a permanent flow of nicotine
5) enough coffee to stay hyper for weeks,
6) a few cans of hertog jan to celebrate the success afterwards
7) a vast determination to follow this guide for setting up RAID in Slackware
http://slacksite.com/slackware/raid.html

should all be pretty straight forward right? i was expecting to get stuck somewhere when trying to set up RAID after the install, since its the first time for me.

but what happens?
1) put in my cd1
2) boot the bare.i
3) login as root
4) cfdisk
error: opened disk read-only, you have no permission to write
then it asks if i want to start with a zero table y/N
when i answer y, it shows me a free space of 667 MB (could this be my cdrom?)

have been trying several versions of:
cfdisk /dev/hda
cfdisk /dev/hdb
cfdisk /etc/etc/etc/etc....

all with no result, so after spending another morning on google, i return here earlier then expected and slightly ashamed cause i have two enormous harddrives and i can't even find one!

anybody?

First we need to know where your hard drive is located in the IDE port

From the CD, try ls /proc/ide
Then cat /proc/ide/hd*/model to verify (eg cat /proc/ide/hda/model for hda)

After that try cfdisk -z /dev/hda (or hdb or whatever from the previous outputs)

oke!!!

in /proc/ide i have the following hda ide0 piix

cat /proc/hda/model gives plextor dvdr (so that explains the 667MB)
cat /proc/ide0/model gives pci
cat /proc/piix/model gives nothing

cfdisk -z /dev/* with all options i can imagine doesn't give aything

Are you sure that your hard drive is IDE ?

From yout output, the system see only one device connected to
IDE port 0 (dvdr drive)