#!/bin/ksh
# $Id: ntp,v 1.8 2008/10/26 18:53:16 ksb Exp $
# Run some checks & report some information on the local host.
#

#****p* netlint/ntp
# NAME
#   ntp
# DESCRIPTION
#   Each host should run a time synchronization daemon (viz. xntpd or ntpd) to
#   keep the system clock accurate. Netlint sanity checks the configuration of
#   these programs.
# IMPACT
#   A host with a bad clock will report errors in the past or future, either
#   of which makes tracking cause and effect hard (at best). Filesystem dumps,
#   log file, cron jobs, and the atq can all be hosed by a poor time-base.
#***

#****a* ntp/STDERR_NTP
# NAME
#   STDERR_NTP
# FORMAT
#   STDERR_NTP: text
# DESCRIPTION
#   The stderr output from ntpdc or xntpdc, or ntpq.
# SEE ALSO
#   STDERR_program
#***

progname=`basename $0`
: ${CARP:=echo}
: ${NTPCONF:=/etc/inet/ntp.conf}
[ ! -f "$NTPCONF" ] && NTPCONF=/etc/inet/ntp.conf
[ ! -f "$NTPCONF" ] && NTPCONF=/etc/ntp.conf
if [ _-V = _$1 ] ; then
	echo "$progname: "'$Id: ntp,v 1.8 2008/10/26 18:53:16 ksb Exp $'
	[ ! -s "$NTPCONF" ] && echo "$progname: $NTPCONF: missing?"
	exit 0
fi
TFILE=/tmp/ntpd.$$.$((RANDOM%100))

#****a* ntp/NTP_STRATUM
# NAME
#   NTP_STRATUM
# FORMAT
#   NTP_STRATUM: stratum
# DESCRIPTION
#   The running ntpd (or xntpd) reported this stratum to netlint. Site policy
#   might decide that this is good or bad. The only really bad value is 16
#   which generally means that we can't find a peer, but  we'll report that
#   elsewhere.
#***

# check our stratum and ntp servers
STRATUM=`echo rv |ntpq |sed -n -e "s/.*stratum=\([0-9]*\).*/\1/p"` 2>/dev/null
: ${STRATUM:=`echo sysinfo | xntpdc | sed -n -e 's/stratum[	: ]*\([0-9]*\)/\1/p'`}
$CARP "NTP_STRATUM: ${STRATUM:-.}"
ntpq -np 2>$TFILE.err >$TFILE.out ||
	xntpdc -np 2>$TFILE.err >$TFILE.out
if [ -s $TFILE.err ]; then
	sed -e 's/^/STDERR_NTP: /' <$TFILE.err
else

#****a* ntp/NTP_PEER_DEAD
# NAME
#   NTP_PEER_DEAD
# FORMAT
#   NTP_PEER_DEAD: nodename
# DESCRIPTION
#   A node which the node tries to use as a time-base is not responding to our
#   polls.
# REMEDIATION
#   Find a better time-base. Also, check that node for issues; it might just be
#   retired.
#***

#****a* ntp/NTP_PEER_SYNC
# NAME
#   NTP_PEER_SYNC
# FORMAT
#   NTP_PEER_SYNC: no server found
# DESCRIPTION
#   None of our time-bases provide any joy.
# REMEDIATION
#   Our clock is more than likely off in anycase. Reset the clock on the host,
#   check the configuration of ntpd and restart ntpd.  /etc/init.d/*ntp* might
#   be the start script.
#***

#****a* ntp/NTP_PEER_FEW
# NAME
#   NTP_PEER_FEW
# FORMAT
#   NTP_PEER_FEW: integer
# DESCRIPTION
#   We don't have enough peers. Site policy should be used here. We hard code
#   that every node should have at least 2 peers for a valid time-base. See the
#   internal NTP document for a tactic to pick new peers.
# REMEDIATION
#   Remediate the configuration and restart the daemon. Check the report on the
#   next major report cycle.
#***

	awk 'BEGIN {
			STAR=0;
			GOODNTP=0
		}
		/^\*/{
			 STAR=1;
		}
		/refid|===/{
			next;
		}
		{
			if ($5 ~ /-/ && $0 ~ /^ /)
				print "NTP_PEER_DEAD: "$1;
			else
				GOODNTP++;
		}
		END {
			if (0 == STAR)
				print "NTP_PEER_SYNC: no server found";
			if (GOODNTP < 2)
				print "NTP_PEER_FEW: ", GOODNTP;
		}
	' <$TFILE.out
fi | xapply -f "$CARP '%1'" -

rm $TFILE.err $TFILE.out
exit 0