Troubleshooting a Hanging Init Script in Solaris 8

Solaris 8 boot scripts (/etc/rc?.d/*) can hang indefinitely with almost no feedback. Below are the methods that actually work on real Solaris 8 systems.

From the ok prompt:

ok boot -s

You land at the root shell (#) without most rc-scripts are not executed yet.

Now manually test the suspect script with full visibility:

# sh -v /etc/rc3.d/SxxProblemScript

or with tracing:

# sh -x /etc/rc3.d/SxxProblemScript 2>&1 | tee /tmp/debug.out

In single-user mode edit /etc/inittab and temporarily replace the initdefault line:

is:3:initdefault:

with

is:3:wait:/sbin/sh < /dev/console > /dev/console 2>&1

Reboot. init will stop after parsing inittab and give you a shell before any rc scripts run.

Then manually execute them one by one:

# for i in /etc/rcS.d/S*; do echo "=== $i ==="; sh -x $i start; done
# for i in /etc/rc2.d/S*; do echo "=== $i ==="; sh -x $i start; done
# for i in /etc/rc3.d/S*; do echo "=== $i ==="; sh -x $i start; done

When one hangs, Ctrl-C and debug it.

In single-user mode edit the suspect script and put at the very top:

exec > /tmp/SxxScript.log 2>&1
set -x
echo "=== Started at $(date) ==="

Reboot normally. When it hangs again, boot boot -s, mount root if needed, and read /tmp/SxxScript.log – it shows exactly where it stopped.

• NIS/ypbind hanging because master/slave is unreachable (ypwhich hangs)
• NFS mounts in /etc/vfstab without bg,intr,soft when server is down
• route add default … when gateway is unreachable
• /etc/rcS.d/S92volmgt starting vold (waits forever if RPC is broken)
• S69inet starting in.routed or in.rdisc
• keyserv, automount, or anything waiting for NIS
• Old Sun Cluster scripts accidentally left enabled

• Boot Solaris 8 CD → single-user
• Mount root disk under /a (usually /dev/dsk/c0t0d0s0)
• Rename or edit the bad script under /a/etc/rc?.d/

# ps -ef | grep -vE "ps|grep"
# tail -50 /var/adm/messages
# grep yp /var/adm/messages
# ypwhich                     # hangs? → NIS problem
# mount -p                    # anything stuck?
# netstat -rn                 # default route present?

99 % of Solaris 8 boot hangs are either NIS/ypbind or an NFS mount without proper timeout options.