Technical article 110825

OpenServer 5.0.5, system hangs just before the login prompt when booting to multiuser mode.

OpenServer 5.0.5

Written · Last revised

Problem

On boot to multiuser mode, the system hangs just before the "login:" prompt is displayed. Telnetting to the system succeeds and the process table shows a sleep 1000000 process running. When the sleep process is killed, the login prompt appears and the system functions normally.

Cause

The sleep command is called by the script /etc/rc2. This is the script that runs when the server goes into multiuser state. The addition of the sleep 1000000 to the script is only present in SCO OpenServer 5.0.5. It addresses some timing problems with drivers starting from this script. With OpenServer 5.0.5, the startup scripts can all start in parallel, making the startup sequence faster. However, a script failure (hang) can completely hang the startup process if one script hangs. With the addition of the sleep command, if a third-party device driver (installed in /etc/idrc.d for example) were to lock, then the processor would still be kept busy running the sleep command until the offending device driver timed out. This prevents the server from completely hanging.

The root cause of this hang is the incomplete startup of a
third-party driver.  The sleep command hanging is a symptom of the
root problem.

Solution

To discover which driver startup script is causing the hang, put
echo statements in the top of each of the startup scripts in
/etc/rc2.d.  The echo statement should echo the name of the script
executing.  The third-party device driver startup script last echoed
before the hang will be the culprit. Typically, this will be a
script from the /etc/idrc.d directory.

Once the offending startup script has been identified, it can be moved to the /etc/rc script so it is started after the system boots to multiuser mode. Also, the script can be examined closely to determine the cause of the hang.

To work around the problem, eliminate the sleep commands from the rc2 script as follows:

Edit the /etc/rc2 script, comment out the lines executing the sleep command:

# PC 6300+ Style Installation-execute rc scripts from driver packages
if [ -d /etc/idrc.d ] then
#
# sleeps only present from 5.0.5 and present for driver timing issues
# Removed SB 28/07/99
#
# sleep 1000000>/dev/null 2>&1 &
# P=$!
{         for f in `ls /etc/idrc.d/* 2>/dev/null`
          do
              if [ -s ${f} ] then
                      /bin/sh ${f}
              fi
          done
          # kill -ALRM $P
}   2>&1 | { $LOGCMD & } 2>/dev/null
# wait $P fi

The less extreme change of eliminating logging can sometimes be enough to address the problem. The results of scripts being run in /etc/idrc.d get recorded in /usr/adm/rc2.log. This write is sometimes the cause of the hang. You can stop this from happening with the following changes to the /etc/rc2 script (adding the '#' before the { $LOGCMD & }):

# PC 6300+ Style Installation - execute rc scripts from driver
packages
if [ -d /etc/idrc.d ] then
      sleep 1000000 >/dev/null 2>&1 &
      P=$!
      {
        for f in `ls /etc/idrc.d/* 2>/dev/null`
        do
              if [ -s  ${f} ] then
                /bin/sh ${f}
              fi
        done
        kill -ALRM $P
      } 2>&1 # | { $LOGCMD & } 2>/dev/null
      wait $P fi
NOTES:    If the problem still remains, telnet to the server during the hang
          and take note of process table contents.  Identify the driver
          startup process that is hanging and add a set -x statement to the
          script. Start it manually to see when the lock happens.

From the Xinuos technical article database, record 110825. Reproduced with permission. The original is at www.sco.com/ta/110825, which is reachable but blocked from search engines.

Next step

Need this fixed rather than explained?

Tell us the release, what the system runs and what it is doing. We will reply by email with what we would do next.

  • You deal with the people who do the work
  • No obligation, and no upgrade required before we will help
  • We tell you plainly if something is not worth doing

Send an enquiry

We reply by email. If a server is down, say so in the first line. You can also write to help@scoexpert.com.