HP-UX — Complete Guide to Enterprise UNIX

HP-UX is Hewlett-Packard’s enterprise UNIX operating system for HP Integrity and PA-RISC servers, featuring the LVM storage stack, Ignite-UX automated provisioning, Serviceguard clustering for 99.999% uptime, and comprehensive system administration tools like SAM and swinstall.

What You’ll Learn & Why It Matters

In this tutorial, you’ll master HP-UX 11i v3’s core capabilities: LVM for flexible storage management, Ignite-UX for network-based OS deployment, Serviceguard for high-availability clustering, and the SAM/swinstall toolchain for day-to-day administration. HP-UX powers the world’s most demanding HP-UX environments — from airline reservation systems to hospital patient databases — where a minute of downtime can mean millions in losses.

Real-world use: Emirates Airlines runs its entire reservation and flight operations on HP-UX. Siemens healthcare uses HP-UX for MRI and CT scan image processing. The London Ambulance Service trusts HP-UX for its emergency dispatch systems. When lives are on the line, HP-UX delivers.

The HP-UX Legacy

HP-UX debuted in 1984 as Hewlett-Packard’s version of AT&T System V UNIX. It has since evolved through three major lineages — on HP 9000 (PA-RISC), HP Integrity (Itanium), and now HP Integrity with x86 compatibility layers.

graph LR
    HPUX1["HP-UX 1.0 (1984)
Series 200/500"] --> HPUX10["HP-UX 10.0 (1995)
SAM, CDE, LVM"]
    HPUX10 --> HPUX11["HP-UX 11.0 (1997)
64-bit, JFS,
Volume Manager"]
    HPUX11 --> HPUX11iv1["HP-UX 11i v1 (2000)
Serviceguard,
Ignite-UX"]
    HPUX11iv1 --> HPUX11iv2["HP-UX 11i v2 (2003)
Itanium support,
Virtual Partitions"]
    HPUX11iv2 --> HPUX11iv3["HP-UX 11i v3 (2007)
Current major release
TSUE, VxFS,
Multi-core support"]

    style HPUX11iv3 fill:#1565C0,color:#fff
    style HPUX11 fill:#FF9800,color:#fff

HP-UX LVM: Storage Management Done Right

HP-UX’s Logical Volume Manager is the foundation of its storage stack. While Linux LVM was inspired by HP-UX LVM, the HP-UX version is more battle-tested and feature-rich in enterprise scenarios.

Physical Volumes, Volume Groups, and Logical Volumes

The hierarchy is familiar if you’ve used any LVM system:

• Physical Volume (PV) — a disk or disk partition
• Volume Group (VG) — a pool of PVs
• Logical Volume (LV) — a virtual disk carved from the VG

HP-UX LVM Commands

# List all volume groups
vgdisplay

# Show details of a specific VG
vgdisplay -v /dev/vg00

# Create a new volume group
mkdir /dev/vgapps
mknod /dev/vgapps/group c 64 0x020000
pvcreate /dev/disk/disk3
vgextend /dev/vgapps /dev/disk/disk3

# Create a logical volume (2GB, contiguous allocation)
lvcreate -n appdata -L 2048 -C y /dev/vgapps

# Create a file system on it
newfs -F vxfs /dev/vgapps/rappdata

# Mount it
mount /dev/vgapps/rappdata /apps

Expected output (vgdisplay -v /dev/vg00):

--- Volume groups ---
VG Name                     /dev/vg00
VG Write Access             read/write
VG Status                   available
Max LV                      256
Cur LV                      12
Open LV                     12
Max PV                      16
Cur PV                      2
Act PV                      2
Max PE per PV               2048
VGDA                        4
PE Size (Mbytes)            4
Total PE                    3560
Alloc PE                    2848
Free PE                     712
Total PVG                   0

Stripping and Mirroring

# Create a striped LV across 3 disks for database performance
lvcreate -n dblv -L 10240 -i 3 -I 64 /dev/vgapps

# Create a mirrored LV with 2 copies
lvcreate -n mitroredlv -L 5120 -m 1 /dev/vgapps

# Convert an existing LV to mirrored on-the-fly
lvextend -m 1 /dev/vgapps/appdata

# Check mirror status
lvdisplay -v /dev/vgapps/appdata | grep "Mirror"

Online File System Growth

One of HP-UX’s hallmark features — grow file systems without unmounting:

# Extend the LV by 500MB
lvextend -L +500 /dev/vgapps/appdata

# Grow the VxFS file system online
fsadm -b 500M /apps

# Verify
bdf /apps

Expected output (bdf /apps):

Filesystem          kbytes    used   avail %used Mounted on
/dev/vgapps/appdata 5242880 1234567 4008313   23% /apps

Serviceguard: High-Availability Clustering

Serviceguard is HP-UX’s clustering solution — it monitors applications, servers, and networks, and automatically fails over services to a healthy node if something fails. It’s designed for 99.999% uptime (five nines), which means less than 5 minutes of downtime per year.

How Serviceguard Works

A Serviceguard cluster has:

• Active nodes — servers running applications
• Standby nodes — servers waiting to take over
• Package — an application + its IP address, volume group, and file system
• Heartbeat — network signals that prove a node is alive

Serviceguard Cluster Configuration

# Check cluster status
cmviewcl -v

# Verify cluster configuration
cmcheckconf -v /etc/cmcluster/cluster.ascii

# Apply configuration
cmapplyconf -v /etc/cmcluster/cluster.ascii

# Start the cluster
cmruncl

# Start a specific package
cmpkg -a myapp_pkg

# View package status
cmviewcl -p myapp_pkg

Expected output (cmviewcl -v):

Cluster:    PROD_CLUSTER
Nodes:      node1 (node2)
State:      up

Package:    myapp_pkg
Node:       node1
State:      running
Node:       node2
State:      up

What Happens During Failover

sequenceDiagram
    participant C as Client
    participant N1 as Node 1 (Primary)
    participant N2 as Node 2 (Standby)
    participant VG as Shared Storage

    N1->>N2: I'm alive (heartbeat)
    N2->>N1: I'm alive (heartbeat)
    C->>N1: Request data
    N1->>C: Response

    Note over N1: Server crashes!
    N2->>N2: Heartbeat timeout (3 missed)
    N2->>VG: Acquire volume group (vgchange -a e)
    N2->>VG: fsck file system
    N2->>N2: Start application
    N2->>VG: Mount file system
    N2->>C: New owner — service restored
    Note over N2: Total failover: 30-90 seconds

Ignite-UX: Network OS Provisioning

Ignite-UX is HP-UX’s answer to Solaris NIM or Linux Kickstart — a network-based system for installing, configuring, and recovering HP-UX machines.

Key Ignite-UX Concepts

Installing a Client with Ignite-UX

# On the Ignite server, create a configuration
cat > /var/opt/ignite/config/client.conf << EOF
hostname = client01
ip_address = 192.168.1.100
netmask = 255.255.255.0
gateway = 192.168.1.1
root_disk = disk0
install_type = hpux.11.31
EOF

# Add software depots
swpackage -s /tmp/hpux-11.31.depot \*
swpackage -s /tmp/patches.depot \*

# Start the Ignite server
/usr/lbin/ignite/ignite -f /var/opt/ignite/config/client.conf

# Boot the client from the network (at client console):
# -> Main Menu -> Boot from LAN

System Recovery with make_archive

# Create a full system archive for disaster recovery
make_archive -a /var/opt/ignite/archives/prodserver.archive -v

# Restore from archive (boot client from Ignite server)
# At the Ignite menu, select "Recover from archive"

System Administration: SAM, swinstall, and parstatus

SAM (System Administration Manager)

SAM is HP-UX’s equivalent of AIX’s SMIT — a menu-driven tool that simplifies complex admin tasks:

# Launch SAM interactively
sam

# Run SAM for a specific area
sam -r storage
sam -r networking
sam -r users
sam -r kernel

SAM groups operations into logical categories: Users and Accounts, Disks and File Systems, Networking, Kernel Configuration, and Performance Monitoring.

swinstall — Software Management

HP-UX uses the Software Distributor (SD) system for package management. It’s more granular than RPM or APT:

# List installed software bundles
swlist

# List a specific bundle's filesets
swlist -l fileset HPUXBase64

# Install software from a depot
swinstall -s /depot/gcc-12.3.0.depot -x mount_all_filesystems=false \*

# Remove software
swremove GCC

# Verify software integrity
swverify GCC

Expected output (swlist):

# Initializing...
# Contacting target "node1"...

Target: node1:/

  HPUXBase64    11.31     HP-UX Base OS (64-bit)
  HPUXBaseAudi  11.31     HP-UX Audio Support
  HPUXBaseNet   11.31     HP-UX Base Networking
  HPUXBaseSec   11.31     HP-UX Base Security
  GCC           12.3.0    GNU Compiler Collection

parstatus — Partition Management

HP-UX on Integrity servers uses hardware partitioning (nPars, vPars) for isolation:

# View all hardware partitions
parstatus

# View details of a specific partition
parstatus -p 0 -v

# Create a virtual partition
vparcreate -p vpar1 -a cpu::1 -a mem::2048 -a io:0::0:0.0.0

# Start a virtual partition
vparboot -p vpar1

Expected output (parstatus):

Partition    Status   CPUs   Memory (MB)   I/O
  0 (nPar)   active    4        8192       3 cells
  1 (nPar)   active    2        4096       1 cell

HP-UX 11i v3 Key Features

HP-UX 11i v3 (the current major release) introduced several enterprise-critical capabilities:

Common Errors & Mistakes

1. Modifying Kernel Parameters Incorrectly

Mistake: Editing /stand/system directly and creating a kernel that won’t boot.

Fix: Always use sam → Kernel Configuration → Configurable Parameters, or run kmtune to change parameters safely. After changes, run mkboot -m to rebuild the kernel. Keep at least one backup kernel in /stand/vmunix.prev.

2. Running Out of Device Special Files

Mistake: Creating many LVs without running insf -e to generate device files, then wondering why mount fails with “no such device.”

Fix: After creating new LVs, always run insf -e to create the corresponding device special files in /dev. Use ls /dev/vg*/r* to verify they exist.

3. Ignoring VGDA and Quorum

Mistake: Losing a disk in a 2-disk volume group and having the VG become unavailable because quorum is lost.

Fix: In Serviceguard environments, set vgchange -Q n on shared volume groups to disable strict quorum enforcement. A storage admin should always understand quorum: with 2 PVs in a VG, losing one disk loses majority — the VG deactivates. With 3 PVs, losing one preserves majority.

4. Forgetting to Run fsck After an Abrupt Shutdown

Mistake: Recovering from a power outage and mounting file systems directly without checking them first.

Fix: Always run fsck -F vxfs -o full on VxFS file systems after an unclean shutdown. For JFS, use fsck -F jfs. Boot to maintenance mode (hpux -lm) and check all file systems before mounting.

5. Not Configuring Serviceguard Heartbeat Over Dedicated Networks

Mistake: Sharing heartbeat traffic with production network traffic, causing false failovers during peak loads.

Fix: Use dedicated network interfaces for Serviceguard heartbeats — ideally two separate physical networks (LAN and RS-232 serial link). Configure cmcld.conf with HEARTBEAT_INTERVAL=1 and NODE_TIMEOUT=6 for faster failure detection.

Practice Questions

Question 1

What are the three layers of HP-UX’s LVM, and what does each do?

Question 2

What is Serviceguard and what problem does it solve?

Question 3

How do you grow a file system in HP-UX without unmounting it?

Question 4

What is Ignite-UX used for?

Question 5

What’s the difference between nPars and vPars?

Challenge

Design and implement a two-node HP-UX Serviceguard cluster:

1. Configure shared storage via LVM with a dedicated volume group for cluster data
2. Set up Serviceguard with heartbeat on two dedicated network links
3. Create a package that includes an Apache web server, its IP address, and its file system
4. Write a test script that induces failover (kill the Apache process) and verifies the service moves to the standby node
5. Configure Ignite-UX to capture and restore the complete configuration as an archive

Real-World Task

Your healthcare organization is deploying a new patient records system on HP-UX Integrity servers. Create the infrastructure plan:

1. Design the LVM layout — separate volume groups for OS, application, and patient data (with mirroring for the data VG)
2. Configure Serviceguard for automatic failover (target: <60 second recovery)
3. Use Ignite-UX to create a gold image of the configured system for rapid deployment
4. Set up SAM-based user administration with role-based groups for doctors, nurses, and administrators
5. Create Bash scripts that monitor critical services and email alerts on failure
6. Document the full disaster recovery procedure using Ignite-UX archives

Mini Project: HP-UX Resource Monitor Script

Write a shell script that collects key system metrics from HP-UX:

#!/bin/sh
# hpux-monitor.sh — HP-UX System Resource Monitor

echo "=========================================="
echo "  HP-UX Resource Monitor"
echo "  Host: $(hostname)"
echo "  OS:  $(uname -r)"
echo "=========================================="

echo ""
echo "--- CPU Partition Info ---"
parstatus 2>/dev/null || echo "No hardware partitions found"

echo ""
echo "--- Memory ---"
echo "Physical: $(dmesg | grep "Physical:" | awk '{print $2}') MB"
echo "Available: $(vmstat -s | grep "available pages" | awk '{print $1*4/1024}') MB"
echo "Active processes: $(ps -ef | wc -l)"

echo ""
echo "--- Volume Groups ---"
vgdisplay | grep -E "VG Name|Total PE|Alloc PE|Free PE"

echo ""
echo "--- Top 5 Large File Systems ---"
bdf | tail -n +2 | sort -k2 -rn | head -5

echo ""
echo "--- Software Patches Installed ---"
swlist -l patch | grep -c "PH" 2>/dev/null
echo "patches currently installed"

echo ""
echo "--- Serviceguard Status ---"
cmviewcl -l 2>/dev/null || echo "Serviceguard not running"

echo ""
echo "=========================================="
echo "  Report generated: $(date)"
echo "=========================================="

Built by the developers of Doda Browser, DodaZIP, and Durga Antivirus Pro.