Fixed issues with the distributed computing interface. Clients upload the status to the server. 'ppss status' does not need to poll all clients for their status.

This commit is contained in:
louwrentius 2011-08-07 11:52:48 +00:00
parent 126e71e231
commit 29713bcc14

67
ppss
View File

@ -1110,8 +1110,9 @@ upload_status () {
set_status () { set_status () {
STATUS="$1" STATUS="$1"
NO_PROCESSED=$(wc -l "$PPSS_HOME_DIR/$LIST_OF_PROCESSED_ITEMS" | cut -d " " -f 1) NO_PROCESSED=$(wc -l "$LIST_OF_PROCESSED_ITEMS" | awk '{ print $1 }' )
NODE=`cat $PPSS_DIR/$NODE_ID` NODE=`cat $PPSS_DIR/$NODE_ID`
echo "$NODE $HOSTNAME $STATUS $NO_PROCESSED" > "$NODE_STATUS_FILE" echo "$NODE $HOSTNAME $STATUS $NO_PROCESSED" > "$NODE_STATUS_FILE"
if [ ! -z "$SSH_SERVER" ] if [ ! -z "$SSH_SERVER" ]
then then
@ -1259,7 +1260,10 @@ deploy () {
ERROR=0 ERROR=0
set_error () { set_error () {
if [ ! "$1" == "0" ] if [ "$ERROR" == "1" ]
then
ERROR=1
elif [ ! "$1" == "0" ]
then then
ERROR=1 ERROR=1
fi fi
@ -1273,7 +1277,9 @@ deploy () {
KEY=`echo $SSH_KEY | cut -d " " -f 2` KEY=`echo $SSH_KEY | cut -d " " -f 2`
ssh -q $SSH_OPTS_NODE $SSH_KEY $USER@$NODE "cd ~ && mkdir -p $PPSS_HOME_DIR && mkdir -p $PPSS_HOME_DIR/$JOB_LOG_DIR && mkdir -p $PPSS_HOME_DIR/ITEM_LOCK_DIR >> /dev/null 2>&1" ssh -q $SSH_OPTS_NODE $SSH_KEY $USER@$NODE "cd ~ && mkdir -p $PPSS_HOME_DIR && mkdir -p $PPSS_HOME_DIR/$JOB_LOG_DIR && mkdir -p $PPSS_HOME_DIR/ITEM_LOCK_DIR >> /dev/null 2>&1"
set_error $?
ssh -q $SSH_OPTS_NODE $SSH_KEY $USER@$NODE "cd ~ && cd $PPSS_HOME_DIR && cd $PPSS_DIR && echo $NODE > $NODE_ID" ssh -q $SSH_OPTS_NODE $SSH_KEY $USER@$NODE "cd ~ && cd $PPSS_HOME_DIR && cd $PPSS_DIR && echo $NODE > $NODE_ID"
set_error $?
scp -q $SSH_OPTS_NODE $SSH_KEY $0 $USER@$NODE:~/$PPSS_HOME_DIR scp -q $SSH_OPTS_NODE $SSH_KEY $0 $USER@$NODE:~/$PPSS_HOME_DIR
set_error $? set_error $?
scp -q $SSH_OPTS_NODE $SSH_KEY $KEY $USER@$NODE:~/$PPSS_HOME_DIR scp -q $SSH_OPTS_NODE $SSH_KEY $KEY $USER@$NODE:~/$PPSS_HOME_DIR
@ -2663,20 +2669,41 @@ start_all_workers () {
#} #}
#get_node_status_from_server () { #get_node_status_from_server () {
get_status_of_node () { get_status_of_nodes () {
RESULT_FILE="$1"
ssh $SSH_OPTS $SSH_KEY $USER@$SSH_SERVER cat "$PPSS_HOME_DIR/$PPSS_NODE_STATUS/*" > "$RESULT_FILE" 2>&1
RES=`ssh $SSH_OPTS $SSH_KEY $USER@$SERVER cat "$PPSS_HOME_DIR/$PPSS_NODE_STATUS/*" 2> /dev/null`
IFS=$'\n' IFS=$'\n'
for x in $RES
for x in `cat $RESULT_FILE`
do do
log DSPLY "$x" IP=`echo $x | awk '{ print $1 }'`
HOST=`echo $x | awk '{ print $2 }'`
STATUS=`echo $x | awk '{ print $3 }'`
RES=`echo $x | awk '{ print $4 }'`
if [ -z "$RES" ]
then
RES="0"
fi
PROCESSED=$((PROCESSED+RES))
LINE=`echo "$IP $HOST $RES $STATUS" | awk '{ printf ("%-16s %-18s % 10s %10s\n",$1,$2,$3,$4) }'`
log DSPLY "$LINE"
done done
log DSPLY "---------------------------------------------------------"
LINE=`echo $PROCESSED | awk '{ printf ("Total processed: % 29s\n",$1) }'`
log DSPLY "$LINE"
rm "$RESULT_FILE"
} }
show_status () { show_status () {
source $CONFIG . $CONFIG
if [ ! -z "$SSH_KEY" ] if [ ! -z "$SSH_KEY" ]
then then
SSH_KEY="-i $SSH_KEY" SSH_KEY="-i $SSH_KEY"
@ -2708,29 +2735,9 @@ show_status () {
log DSPLY "---------------------------------------------------------" log DSPLY "---------------------------------------------------------"
PROCESSED=0 PROCESSED=0
for x in `cat $NODES_FILE` RESULT_FILE="$RADOM$RANDOM.deleteme"
do get_status_of_nodes "$RESULT_FILE"
RES=0
NODE=`get_status_of_node "$x" | awk '{ print $1 }'`
if [ ! "$NODE" == "UNKNOWN" ]
then
STATUS=`get_status_of_node "$x" | awk '{ print $2 }'`
RES=`exec_cmd "grep -i $NODE ~/$PPSS_HOME_DIR/$JOB_LOG_DIR/* 2>/dev/null | wc -l " 1`
if [ ! "$?" ] || [ -z "$RES" ]
then
RES=0
fi
else
STATUS="UNKNOWN"
RES=0
fi
let PROCESSED="$PROCESSED+$RES"
LINE=`echo "$x $NODE $RES $STATUS" | awk '{ printf ("%-16s %-18s % 10s %10s\n",$1,$2,$3,$4) }'`
log DSPLY "$LINE"
done
log DSPLY "---------------------------------------------------------"
LINE=`echo $PROCESSED | awk '{ printf ("Total processed: % 29s\n",$1) }'`
log DSPLY "$LINE"
} }
main () { main () {