Disk allocation for HPUX 11.00

[Ray Carlson]

I just got a N4000 with 4GB of Ram, 2 internal 18GB disks, and 10 external drives (2-72GB, 2-36GB, & 6-18GB). I plan to install HPux 11.00 64-bit, Mirror-UX, and Oracle 8.1.7 on this machine. Any recommendations for system partition sizes, swap space, disk layouts extra? I will be mirroring the system disk, but not sure about the data disks since the Oracle database already takes 98GB and is expected to grow even larger. Thanks in advance for any help or suggestions.

[Pete Randall]

Hi Ray,

Attached is our layout for the root volume group. We have a very similar layout: N4000, 8Gb, 2x18Gb disks, and externals for the DB, which is Informix in our case but should be similar to your situation.

I should explain that our /var is vastly oversized because we store huge quantities of print output.

HTH,
Pete

[Leif Halvarsson_2]

Hi
I have never liked the idea to divide the OS installation in several small logical volumes (/ /stand /var /opt etc.)as default when installing HP-UX. I usually create one small hfs volume for /stand (as this is necessary) and one large (4GB) for / (rest of the OS). Then I create logical volumes for data applications etc. as needed.

I think it is much more important to have som kind of security (mirror, raid) for the Oracle database then for the OS. Restoring the OS disk can be done in a few minutes if using a Ignite server and very fast if using a Ignite tape. But restoring a large Oracle database can be a very hard work.

[Ravi_8]

Hi,

we have N4000 with 5 external disks, i am using for Virtual partitions( 11i)

[ASO CENTRAL]

For small workstations (single user), one large root partition is probably OK. However, the reliability and resilience of a large systems (multiple users, esp. users with shell access) is significantly improved by separating filesystems and subdividing /var even further.

In a large / filesystem, a single rogue email with multi-gigabyte crash dump will fill / and down you go. A user can't print and rather than see what's wrong, will print again (and again and again...) filling /var/spool. An application goes berserk and creates massive files in /tmp or /var/tmp. A user is learning Unix for Dummies and tries the prealloc command filling /home. The system panics due to a missing patch and fills /var/adm/crash with crash dumps...(and so on)

Each scenario is catastrophic to a single / filesystem, yet survivable with separate mount points. /home is self-evident...users will be abusers so keep them out of the HP-UX mountpoints. /tmp should be carefully managed although SysV.4 standards recommend /var/tmp rather than /tmp for all temporary user and application files (about a 10 year old standard). /var should be broken up as dictated by actual usage:

/var
/var/mail (and put limits in sendmail.cf)
/var/adm
/var/adm/crash
/var/adm/sw
/var/spool

Now, a user that prints without thinking may fill /var/spool but at least the applications still run and other users can access the system. Or a panic loop will fill up /var/adm/crash but the system run just fine.

[Leif Halvarsson_2]

Hi
I never have user data (if there is ordinary users on this system) on the root file system, all ordinary users is on a NFS-mounted filesystem. And if I install applications I create a separate filesystem for this (and for application data if needed).
I know there IS a risk the root file system becomes full (for example with logfiles) but this has never happens to me.
But, when using the default filesystem layout I have often had problems with one of this small filesystems become full (for example when installing patches or OS extensions) but there is plenty of space on other filesystems.