rc_pmcd

#!/bin/sh
#
# Copyright (c) 2013-2015,2019 Red Hat.
# Copyright (c) 2000-2008 Silicon Graphics, Inc.  All Rights Reserved.
# 
# This program is free software; you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
# Free Software Foundation; either version 2 of the License, or (at your
# option) any later version.
# 
# This program is distributed in the hope that it will be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
# or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
# for more details.
# 
# Start or Stop the Performance Co-Pilot Collection Daemon (PMCD)
#
# The following is for chkconfig on RedHat based systems
# chkconfig: 2345 95 05
# description: pmcd is the collection daemon for the Performance Co-Pilot (PCP)
#
# The following is for insserv(1) based systems,
# e.g. SuSE, where chkconfig is a perl script.
### BEGIN INIT INFO
# Provides:          pmcd
# Required-Start:    $local_fs
# Should-Start:      $network $remote_fs $syslog $time
# Required-Stop:     $local_fs
# Should-Stop:       $network $remote_fs $syslog
# Default-Start:     2 3 4 5
# Default-Stop:      0 1 6
# Short-Description: Control pmcd (the collection daemon for PCP)
# Description:       Configure and control pmcd (the collection daemon for the Performance Co-Pilot)
### END INIT INFO
#
# For FreeBSD
# PROVIDE: pmcd
# REQUIRE: NETWORKING FILESYSTEMS syslogd
# KEYWORD: shutdown
# And add the following lines to /etc/rc.conf to run pmcd:
# pmcd_enable="YES"
#

. $PCP_DIR/etc/pcp.env
. $PCP_SHARE_DIR/lib/rc-proc.sh

# for chasing arguments we're passed from init/systemd/...
#
#debug# echo "$*: `date`" >>$PCP_LOG_DIR/rc_pmcd.log
#debug# env >>$PCP_LOG_DIR/rc_pmcd.log

PMCD=$PCP_BINADM_DIR/pmcd
PMCDOPTS=$PCP_PMCDOPTIONS_PATH
PCPLOCAL=$PCP_PMCDRCLOCAL_PATH
PMCDENVS=$PCP_SYSCONFIG_DIR/pmcd
RUNDIR=$PCP_LOG_DIR/pmcd
prog=$PCP_RC_DIR/`basename $0`

tmp=`mktemp -d "$PCP_TMPFILE_DIR/pmcd_rc.XXXXXXXXX"` || exit 1
status=1
trap "rm -rf $tmp; exit \$status" 0 1 2 3 15

# optional begin logging to $PCP_LOG_DIR/NOTICES
#
if $PCP_LOG_RC_SCRIPTS
then
    logmsg="begin pid:$$ $prog args:$*"
    if which pstree >/dev/null 2>&1
    then
	logmsg="$logmsg [`pstree -lps $$`]"
	logmsg="`echo "$logmsg" | sed -e 's/---pstree([^)]*)//'`"
    fi
    $PCP_BINADM_DIR/pmpost "$logmsg"
fi

case "$PCP_PLATFORM"
in
    mingw)
	# nothing we can usefully do here, skip the test
	#
	IAM=0
	;;

    *)
	# standard Unix/Linux style test
	#
	ID=id
	IAM=`$ID -u 2>/dev/null`
	if [ -z "$IAM" ]
	then
	    # do it the hardway
	    #
	    IAM=`$ID | sed -e 's/.*uid=//' -e 's/(.*//'`
	fi
	;;
esac

_pmcd_logfile()
{
default=$RUNDIR/pmcd.log
$PCP_AWK_PROG <$PMCDOPTS '
BEGIN		{ logf = "'$default'" }
$1 == "-l"	{ if (NF > 1) logf = $2 }
END		{ print logf }'
}

_reboot_setup()
{
    # base directories and house-keeping for daemon processes
    #
    # For most packages, $PCP_RUN_DIR is included in the package,
    # but for Debian and cases where /var/run is a mounted filesystem
    # it may not exist, so create it here before it is used to create
    # any pid/lock files
    #
    # $PCP_RUN_DIR creation is also done in pmlogger_daily, but if pmcd
    # is running, we'd expect do this one first
    #
    if [ ! -d "$PCP_RUN_DIR" ]
    then
	mkdir -p -m 775 "$PCP_RUN_DIR"
	chown $PCP_USER:$PCP_GROUP "$PCP_RUN_DIR"
	if which restorecon >/dev/null 2>&1
	then
	    restorecon -r "$PCP_RUN_DIR"
	fi
    fi

    # setup and clean up base directories and house-keeping for tracking
    # pmlogger instances ... needs to be done here because pmcd needs the
    # information, even if no pmlogger instances are started (and even if
    # $PCP_RC_DIR/pmlogger is not run)
    #
    if [ ! -d "$PCP_TMP_DIR/pmlogger" ]
    then
	mkdir -p -m 775 "$PCP_TMP_DIR/pmlogger"
	chown $PCP_USER:$PCP_GROUP "$PCP_TMP_DIR/pmlogger"
	if which restorecon >/dev/null 2>&1
	then
	    restorecon -r "$PCP_TMP_DIR"
	fi
    else
	rm -rf $tmp/ent $tmp/pid
	here=`pwd`
	cd "$PCP_TMP_DIR/pmlogger"
	_get_pids_by_name pmlogger | sort >$tmp/pid
	ls [0-9]* 2>&1 | sed -e '/\[0-9]\*/d' \
	| sed -e 's/[ 	][ 	]*//g' | sort >$tmp/ent
	# remove entries without a pmlogger process
	#
	rm -f `comm -23 $tmp/ent $tmp/pid`
	rm -f $tmp/ent $tmp/pid
	# if primary is now a broken symlink, cleanup (-e follows symlinks)
	#
	[ ! -e primary ] && rm -f primary
	cd "$here"
    fi

    # Rebuild PMNS?
    #
    PMNSDIR=$PCP_VAR_DIR/pmns

    rebuild=false
    if [ -d $PMNSDIR -a \( -f $PMNSDIR/.NeedRebuild -o ! -f $PMNSDIR/root \) ]
    then
	rebuild=true
    else
	num=`find $PMNSDIR -newer $PMNSDIR/root -iname 'root_*' 2>/dev/null | wc -l`
	[ "$num" -gt 0 ] && rebuild=true
    fi

    if $rebuild
    then
	if [ -x $PMNSDIR/Rebuild ]
	then
	    $ECHO $PCP_ECHO_N "Rebuilding PMNS ..." "$PCP_ECHO_C"
	    here=`pwd`
	    cd $PMNSDIR
	    ./Rebuild -u $REBUILDOPT
	    $RC_STATUS -v
	    # The 'root' file does not get updated when data did not change,
	    # so we must touch it to update date.
	    [ $? -eq 0 ] && { rm -f .NeedRebuild; touch root; }
	    cd "$here"
	fi
    fi
}

# if _pmda_setup() is interrupted, we need to cleanup
# (a) remove the mutual exclusion lock file, and
# (b) make sure any pending operation will be repeated at the
#     next restart
#
_pmda_setup_cleanup()
{
    echo "_pmda_setup: Interrupted!"
    rm -rf $tmp $PCP_TMPFILE_DIR/pmda.auto.update
    for file in $PCP_PMDAS_DIR/*/.Need*.pending
    do
	[ "$file" = "$PCP_PMDAS_DIR/*/.Need*.pending" ] && continue
	guard=`echo $file | sed -e 's@.*/@@' -e 's/\.pending$//'`
	pmda=`echo $file | sed -e 's@/.Need.*@@' -e 's@.*/@@'`
	echo "_pmda_setup_cleanup: reset $guard for $pmda PMDA"
	mv -f $PCP_PMDAS_DIR/$pmda/$guard.pending $PCP_PMDAS_DIR/$pmda/$guard
    done
}

# Note: _pmda_setup() is run only once per rc script invocation ...
#       even though this may call Install and/or Remove scripts
#       for multiple PMDAs which may in turn (recursively) call
#       the rc script
#
_pmda_setup()
{
    # Auto-Install and/or Auto-Remove PMDAs?
    #
    if $PCP_BINADM_DIR/pmlock $PCP_TMPFILE_DIR/pmda.auto.update
    then
	# we got the lock, proceed
	$PCP_BINADM_DIR/pmpost "PMDA setup: pid=$$ acquire lock"
	:
    else
	# one of us is already running
	#
	$PCP_BINADM_DIR/pmpost "PMDA setup: pid=$$ lock failed"
	return
    fi

    # if we are interrupted, remove the lock file
    trap "_pmda_setup_cleanup; exit \$status" 1 2 3 15

    if [ -d "$PCP_PMDAS_DIR" ]
    then
	here=`pwd`
	cd "$PCP_PMDAS_DIR"
	_pmda_prepare Upgrade
	_pmda_enact Remove Removing
	_pmda_enact Install Installing
	cd "$here"
    fi

    rm -f $PCP_TMPFILE_DIR/pmda.auto.update
    $PCP_BINADM_DIR/pmpost "PMDA setup: pid=$$ release lock"

    # re-instate mainline trap
    #
    trap "rm -rf $tmp; exit \$status" 0 1 2 3 15
}

# Take automated setup action (Install or Remove) for PMDAs
# Protocol here is:
#	.Need${action} => package installs may leave these behind
#			  to indicate action needed here
#	.Need${action}.pending => action is running
#	.Need${action}.failed => action run but failed (these will
#				 trigger a retry the next time pmcd is
#				 restated
#
_pmda_enact()
{
    action=$1
    verb=$2

    for file in */.Need${action}*
    do
	# skip all of this if no files from glob expansion
	#
	[ "$file" = "*/.Need${action}*" ] && break
	pmda=`dirname $file`
	# sanity ... $pmda really should be a directory name
	#
	[ -d "$pmda" ] || continue
	# rename marker file asap and _before_ calling action script
	# because that script might call this start script recursively,
	# and we don't want to get stuck in an infinite loop.
	# 
	case "$file"
	in
	    */.Need${action}|*/.Need${action}.failed|*/.Need${action}.pending)
		case "$file"
		in
		    */.Need${action}.pending)
			# interrupted? we have the mutex lock here, so
			# just proceed
			#
			;;
		    *)
			rm -f $pmda/.Need${action}.pending
			mv $file $pmda/.Need${action}.pending
			;;
		esac
		$PCP_ECHO_PROG "$verb $pmda PMDA ..."
		cd $pmda
		if ./${action} </dev/null >/dev/null
		then
		    # success
		    $PCP_BINADM_DIR/pmpost "PMDA setup: automated $action: $pmda"
		    rm -f .Need${action}.pending
		else
		    # failure
		    $PCP_BINADM_DIR/pmpost "PMDA setup: automated $action FAILED (exit=$?): $pmda"
		    # leave the marker, but rename to .failed, this
		    # will allow another attempt at the next restart
		    # and maybe we'll be luckier next time
		    #
		    mv .Need${action}.pending .Need${action}.failed
		fi
		cd "$PCP_PMDAS_DIR"
		;;

	    *)
		$PCP_BINADM_DIR/pmpost "PMDA automated setup: botch: what is $file?"
		;;
	esac

    done
}

# Take automated corrective steps (Upgrade) for installed PMDAs
#
_pmda_prepare()
{
    action=$1

    for file in */$action
    do
	[ "$file" = "*/$action" ] && break
	pmda=`dirname $file`
	if [ -d "$pmda" -a -f "$pmda/$action" ]
	then
	    cd $pmda
	    if ! ./${action} </dev/null >/dev/null
	    then
		$PCP_BINADM_DIR/pmpost "PMDA setup: automated $action FAILED (exit=$?): $pmda"
	    fi
	    cd "$PCP_PMDAS_DIR"
	fi
    done
}

_start_pmcheck()
{
    if [ ! -z "$PMCD_WAIT_TIMEOUT" ]
    then
	wait_option="-t $PMCD_WAIT_TIMEOUT" 
    else 
	wait_option=''
    fi

    # check to see if pmcd started with explicit Unix domain socket;
    # if so, we'll wait on that interface.
    #
    if [ -z "$PMCD_SOCKET" ]
    then
	sock_option=`sed -n <$PMCDOPTS 2>/dev/null \
			 -e '/^[^-]/d' \
			 -e 's/  / /' \
			 -e 's/^/ /' \
			 -e 's/ -s *//gp'`
	[ -n "$sock_option" ] && export PMCD_SOCKET="$sock_option"
    fi

    if $PCP_BINADM_DIR/pmcd_wait $wait_option
    then
	:
    else
	status=$?
	message="pmcd_wait failed in $prog: exit status: $status"
	$PCP_SYSLOG_PROG -p daemon.warning "$message"
	$PCP_BINADM_DIR/pmpost "$message"
	$RC_STATUS -v
	exit
    fi
}

# Use $PCP_PMCDCONF_PATH to find and terminate pipe/socket PMDAs.
#
_killpmdas()
{
    if [ ! -f $PCP_PMCDCONF_PATH ]
    then
	echo "$prog:"'
Warning: pmcd control file '"$PCP_PMCDCONF_PATH"' is missing, cannot identify PMDAs
         to be terminated.'
	return
    fi

    # First join up continued lines in config file, then find processes
    $PCP_AWK_PROG < $PCP_PMCDCONF_PATH '
/\\\\$/		{ printf "%s ", substr($0, 0, length($0) - 1); next }
		{ print }' \
| $PCP_AWK_PROG '
$1 ~ /^#/				{ next }
tolower($3) == "pipe" && NF > 4		{ $1=$2=$3=$4=""; print $0; next }
tolower($3) == "socket" && NF > 5	{ $1=$2=$3=$4=$5=""; print $0; next }
' > $tmp/pmdas

    cat $tmp/pmdas | while read pmda
    do
	_get_pids_by_name -a "$pmda" >$tmp/tmp
	[ -s $tmp/tmp ] || continue
	pmsignal -s TERM `cat $tmp/tmp` > /dev/null 2>&1 &
    done
    wait

    # Give PMDAs some time after SIGTERM to exit cleanly, then SIGKILL
    pmsleep 0.1

    cat $tmp/pmdas | while read pmda
    do
	_get_pids_by_name -a "$pmda" >$tmp/tmp
	[ -s $tmp/tmp ] || continue
	pmsignal -s KILL `cat $tmp/tmp` > /dev/null 2>&1 &
    done
    wait
}

_shutdown()
{
    # Is pmcd running?
    #
    _get_pids_by_name pmcd >$tmp/tmp
    if [ ! -s $tmp/tmp ]
    then
 	[ "$1" = verbose ] && echo "$prog: pmcd not running"
	rm -f $PCP_RUN_DIR/pmcd.pid $PCP_RUN_DIR/pmcd.socket
	return 0
    fi
    pidlist="`cat $tmp/tmp`"
    suffix=''
    if [ "`wc -l <$tmp/tmp | sed -e 's/ //g'`" -ne 1 ]
    then
 	[ "$1" = verbose ] && echo "$prog: warning more than one pmcd running: $pidlist"
	suffix='es'
    fi

    # If pmcd is running but we can't find a pidfile, or a logfile at the
    # configured or default location, assume this script is being run via
    # a chroot build environment (and hence we do not want to signal pmcd).
    #
    logf=`_pmcd_logfile`
    [ -f $logf ] || logf=$RUNDIR/pmcd.log
    if [ ! -f $PCP_RUN_DIR/pmcd.pid -a ! -f $logf ]
    then
	echo "PMCD process$suffix ... $pidlist"
        echo "$prog:
Warning: found no $PCP_RUN_DIR/pmcd.pid
         and no $logf.
         Assuming an uninstall from a chroot: pmcd not killed.
         If this is incorrect, \"pmsignal -s TERM $pidlist\"  can be used."
        exit
    elif [ -f $PCP_RUN_DIR/pmcd.pid ]
    then
	runpid=`cat $PCP_RUN_DIR/pmcd.pid`
	for pid in $pidlist
	do
	    if [ "$runpid" != "$pid" ]
	    then
		echo "PMCD process ... $pid"
		echo "$prog:
Warning: process ID in $PCP_RUN_DIR/pmcd.pid ($runpid) is different.
         Check logfile $logf. When you are ready to proceed,
	 remove $PCP_RUN_DIR/pmcd.pid before retrying."
		exit
	    fi
	done
    else
	# no $PCP_RUN_DIR/pmcd.pid ... use $pidlist from _get_pids_by_name()
	#
	runpid=''
    fi

    # Send pmcd a SIGTERM, which is noted as a pending shutdown.
    # When finished the currently active request, pmcd will close any
    # connections, wait for any agents, and then exit.
    # On failure, resort to SIGKILL.
    #
    [ "$1" = quietly ] ||\
    $ECHO $PCP_ECHO_N "Waiting for pmcd to terminate ...""$PCP_ECHO_C"
    delay=200	# tenths of a second
    for SIG in TERM KILL
    do
	if [ -z "$runpid" ]
	then
	    pmsignal -s $SIG $pidlist > /dev/null 2>&1
	else
	    pmsignal -s $SIG $runpid >/dev/null 2>&1
	    rm -f $PCP_RUN_DIR/pmcd.pid $PCP_RUN_DIR/pmcd.socket
	fi
	while [ $delay -gt 0 ]
	do
	    _get_pids_by_name pmcd >$tmp/tmp
	    [ ! -s $tmp/tmp ] && break 2
	    pmsleep 0.1
	    delay=`expr $delay - 1`
	    [ "$SIG" = "TERM" ] && [ `expr $delay % 10` -eq 0 ] \
		&& $ECHO $PCP_ECHO_N ".""$PCP_ECHO_C"
	done
	echo
	echo "Process ..."
	if [ "$SIG" = "TERM" ]
	then
	    $PCP_PS_PROG $PCP_PS_ALL_FLAGS >$tmp/ps
	    sed 1q $tmp/ps
	    for pid in `cat $tmp/tmp`
	    do
	        $PCP_AWK_PROG <$tmp/ps "\$2 == $pid { print }"
	    done
	    echo "$prog: Warning: Forcing pmcd to terminate!"
	    delay=20
	else
	    cat $tmp/tmp
	    echo "$prog: Warning: pmcd won't die!"
	    exit
	fi
    done
    _killpmdas

    if [ "$1" = quietly ]
    then
	true && $RC_STATUS
    else
	true && $RC_STATUS -v
    fi
    $PCP_BINADM_DIR/pmpost "stop pmcd from $prog"
}

_usage()
{
    echo "Usage: $prog [-v] {start|start-systemd|faststart|restart|condrestart|stop|stop-systemd|status|reload|force-reload}"
}

VERBOSE_CTL=on
while getopts v c
do
    case $c
    in
	v)  # force verbose ... for historical reasons only as $VERBOSE_CTL
	    # is always "on"
	    ;;
	
	*)
	    _usage
	    exit 1
	    ;;
    esac
done
shift `expr $OPTIND - 1`

if [ $VERBOSE_CTL = on ]
then				# For a verbose startup and shutdown
    ECHO=$PCP_ECHO_PROG
    REBUILDOPT=''
    VFLAG='-v'
else				# For a quiet startup and shutdown
    ECHO=:
    REBUILDOPT=-s
    VFLAG=
fi

if [ "$IAM" != 0 -a "$1" != "status" ]
then
    if [ -n "$PCP_DIR" ]
    then
	: running in a non-default installation, do not need to be root
    else
	echo "$prog:"'
Error: You must be root (uid 0) to start or stop the Performance Co-Pilot pmcd.'
	exit
    fi
fi

# First reset status of this service
$RC_RESET
 
# Return values acc. to LSB for all commands but status:
# 0 - success
# 1 - misc error
# 2 - invalid or excess args
# 3 - unimplemented feature (e.g. reload)
# 4 - insufficient privilege
# 5 - program not installed
# 6 - program not configured
#
# Note that starting an already running service, stopping
# or restarting a not-running service as well as the restart
# with force-reload (in case signalling is not supported) are
# considered a success.
case "$1" in

  start|start-systemd|faststart|restart|condrestart|reload|force-reload)
	if [ "$1" = "condrestart" ] && ! is_chkconfig_on pmcd
	then
	    status=0
	    exit
	fi
	_shutdown quietly

	# Clean the environment for PMCD:
	# PMCD and PMDA messages should go to stderr, not the GUI notifiers
	# Clients in these scripts should test PMCD status without TLS/SSL.
	#
	unset PCP_STDERR PCP_SECURE_SOCKETS

	_reboot_setup

	if [ -x $PMCD ]
	then
	    if [ ! -f $PCP_PMCDCONF_PATH ]
	    then
		echo "$prog:"'
Error: pmcd control file '"$PCP_PMCDCONF_PATH"' is missing, cannot start pmcd.'
		exit
	    fi
	    if [ ! -d "$RUNDIR" ]
	    then
		mkdir -p -m 775 "$RUNDIR"
		chown $PCP_USER:$PCP_GROUP "$RUNDIR"
		if which restorecon >/dev/null 2>&1
		then
		    restorecon -r "$RUNDIR"
		fi
	    fi
	    cd "$RUNDIR"

	    # salvage the previous versions of any PMCD and PMDA logfiles
	    #
	    for log in pmcd `sed -e '/^#/d' -e '/\[access\]/q' -e 's/[ 	].*//' <$PCP_PMCDCONF_PATH`
	    do
		if [ -f $log.log ]
		then
		    rm -f $log.log.prev
		    mv $log.log $log.log.prev
		fi
	    done

	    [ "$1" = "start-systemd" ] ||\
	    $ECHO $PCP_ECHO_N "Starting pmcd ..." "$PCP_ECHO_C"

	    # only consider variables which start with PMCD or PCP
	    `$PCP_AWK_PROG '/^PMCD/||/^PCP/ {print("export", $0)}' < $PMCDENVS 2>/dev/null`

	    # only consider lines which start with a hyphen
	    # get rid of the -f option
	    # ensure multiple lines concat onto 1 line
	    OPTS=`sed <$PMCDOPTS 2>/dev/null \
			    -e '/^[^-]/d' \
			    -e 's/^/ /' \
			    -e 's/$/ /' \
			    -e 's/ -f / /g' \
			    -e 's/^ //' \
			    -e 's/ $//' \
		    | tr '\012' ' ' `

	    if ! $PMCD --verify $OPTS
	    then
		echo $prog: pmcd --verify $OPTS failed, cannot start pmcd.
		exit
	    fi
	    $PMCD $OPTS
            _start_pmcheck
	    $RC_STATUS -v

	    # auto-install and/or auto-removal of PMDAs may trigger
	    # a recursive call into this script to restart pmcd,
	    # so be careful to only run _pmda_setup once and do other
	    # tasks related to the initial script invocation only once
	    #
	    if [ -z "$PMDA_SETUP_RUNNING" ]
	    then
		# not a recursive call ...

		$PCP_BINADM_DIR/pmpost "start pmcd from $prog"

		# site-local customisations after PMCD startup
		[ -x $PCPLOCAL ] && $PCPLOCAL $VFLAG start

		# auto install/remove any PMDAs ... this may trigger the
		# recursion and setting $PMDA_SETUP_RUNNING in the
		# environment is the guard
		#
		PMDA_SETUP_RUNNING=yes _pmda_setup &
	    else
		$PCP_BINADM_DIR/pmpost "start pmcd from PMDA setup called from $prog"
	    fi

	fi
	status=0
        ;;

  stop|stop-systemd)
	# site-local customisations before pmcd shutdown
	#
	[ -x $PCPLOCAL ] && $PCPLOCAL $VFLAG stop
	if [ "$1" = stop-systemd ]
	then
	    _shutdown quietly
	else
	    _shutdown verbose
	fi
	status=0
        ;;

  status)
	# NOTE: $RC_CHECKPROC returns LSB compliant status values.
	$ECHO $PCP_ECHO_N "Checking for pmcd:" "$PCP_ECHO_C"
        if [ -r /etc/rc.status ]
        then
            # SuSE
            $RC_CHECKPROC $PMCD
            $RC_STATUS -v
            status=$?
        else
            # not SuSE
            $RC_CHECKPROC $PMCD
            status=$?
            if [ $status -eq 0 ]
            then
                $ECHO running
            else
                $ECHO stopped
            fi
        fi
	;;

  *)
	_usage
        ;;
esac

# optional end logging to $PCP_LOG_DIR/NOTICES
#
if $PCP_LOG_RC_SCRIPTS
then
    $PCP_BINADM_DIR/pmpost "end pid:$$ $prog status=$status"
fi

exit