core_watch

#############################################################
# script to watch for cores during QA runs, so they won't overwrite one
# another
# Not activated for efficiency reasons, and problems on MKS, us
# only when needed and remember to remove afterwards
#############################################################

#############################################################
# to activate put the following into all.sh (after the HOSTDIR 
# has been exported
#############################################################
# sh `dirname $0`/core_watch $HOSTDIR ${HOSTDIR} &    
# CORE_WATCH_PID=$!
# if [ -n "${KILLPIDS}" ]
# then
#     echo $CORE_WATCH_PID >>"${KILLPIDS}"            
# fi
#############################################################

#############################################################
# or put the following into nssqa to watch the whole RESULTDIR
# start it shortly before run_all
#
# NOTE: the more efficient way is above, this is potentially going
# thru 1000ds of files every 30 seconds
#############################################################
# sh `dirname $0`/core_watch $RESULTDIR &    
# echo $! >>"${KILLPIDS}"        #so Exit() can hopefully kill the core_watch
#############################################################

# in both cases remember to kill the process when done, since 
# the PIDs that end up in ${KILLPIDS} might not work for all OS
# something like "kill_by_name core_watch

echo $$ >>"${KILLPIDS}"     #so Exit() can hopefully kill this shell
while [ 1 ]
do
    for w in `find $1 -name "core" -print`
    do
        echo "Found core $w"
        mv $w $w.`date +%H%M%S`
    done
    sleep 30
done