backup

ppss

@@ -26,7 +26,7 @@
 trap 'kill_process' SIGINT 
 
 SCRIPT_NAME="Distributed Parallel Processing Shell Script"
-SCRIPT_VERSION="2.81"
+SCRIPT_VERSION="2.83"
 
 #
 # The first argument to this script can be a mode.
@@ -89,6 +89,7 @@ ACTIVE_WORKERS="0"
 DAEMON_POLLING_INTERVAL="10"
 STAT=""
 DAEMON_FILE_AGE="4"
+ENABLE_INPUT_LOCK="0"
 
 SSH_SERVER=""                           # Remote server or 'master'.
 SSH_KEY=""                              # SSH key for ssh account.
@@ -118,6 +119,7 @@ ITEM_ESCAPED=""
 DISABLE_SKIPPING=0
 NODE_STATUS="$PPSS_DIR/status.txt"
 DAEMON=0
+EMAIL=""
 
 REGISTER=""                               # For STACK
 STACK=""
@@ -136,6 +138,7 @@ showusage_short () {
     echo "                 $0 -d /dir/with/some/files -c 'gzip '" 
     echo "                 $0 -d /dir/with/some/files -c 'cp \"\$ITEM\" /tmp' -p 2" 
     echo "                 $0 -f <file> -c 'wget -q -P /destination/directory \"\$ITEM\"' -p 10" 
+    echo 
 }
 
 showusage_normal () {
@@ -178,15 +181,19 @@ showusage_normal () {
     echo -e "--quiet | -q       Shows no output except for a progress indication using percents."
     echo
     echo -e "--delay | -D       Adds an initial random delay to the start of all parallel jobs to spread"
-    echo -e "                   the load. The delay is only used at the start of all 'threads'."
+    echo -e "                   the load. The delay (seconds) is only used at the start of all 'threads'."
     echo
-    echo -e "--daemon           Do not exit after items are professed, but keep looking for new items"
+    echo -e "--daemon           Daemon mode. Do not exit after items are professed, but keep looking "
-    echo -e "                   and process them. Read the manual how to use this!"
+    echo -e "                   for new items and process them. Read the manual how to use this!"
+    echo -e "                   See --help for important additional options regarding daemon mode."
     echo
     echo -e "--no-recursion|-r  By default, recursion of directories is enabled when the -d option is "
     echo -e "                   used. If this is not prefered, this can be disabled with this option "
     echo -e "                   Only files within the specified directory will be processed."
     echo 
+    echo -e "--email | -e       PPSS sends an e-mail if PPSS has finished. It is also used if processing"
+    echo -e "                   of an item has failed (configurable, see -h). "
+    echo 
     echo -e "--help             Extended help, including options for distributed mode and Amazon EC2."
     echo 
     echo -e "Example: encoding some wav files to mp3 using lame:"
@@ -254,6 +261,24 @@ showusage_long () {
     echo -e "--daemon           Do not exit after items are professed, but keep looking for new items"
     echo -e "                   and process them. Read the manual how to use this!"
     echo
+    echo -e "--interval         Specifies the polling interval when running in daemon mode. Polls every"
+    echo -e "                   x seconds for new items to process."
+    echo
+    echo -e "--file-age         When not using inotify, specify how many seconds must have passed before"
+    echo -e "                   a file may be processed to prevent files being processed while being "
+    echo -e "                   written to."
+    echo
+    echo -e "--disable-inotify  If for some reason, inotify must not be used, use this option to disable"
+    echo -e "                    usage of inotify. Regular polling will be used."
+    echo 
+    echo -e "--enable-input-lock     When PPSS is run in daemon mode, create a directory INPUT_LOCK to"    
+    echo -e "                   signal that items are processed and may not be touched by PPSS."
+    echo -e "                   Once this directory is removed, PPSS will start processing items."
+    echo
+    echo -e "--no-recursion|-r  By default, recursion of directories is enabled when the -d option is "
+    echo -e "                   used. If this is not prefered, this can be disabled with this option "
+    echo -e "                   Only files within the specified directory will be processed."
+    echo
     echo -e "--no-recursion|-r  By default, recursion of directories is enabled when the -d option is "
     echo -e "                   used. If this is not prefered, this can be disabled with this option."
     echo -e "                   Only files within the specified directory will be processed."
@@ -449,7 +474,6 @@ is_var_empty () {
     fi
 }
 
-
 detect_inotify () {
 
     if [ -e /usr/bin/inotifywait ] && [ ! "$INOTIFY" = "0" ]
@@ -546,6 +570,10 @@ process_arguments () {
                             INOTIFY=0
                             add_var_to_config INOTIFY "$INOTIFY"
                             shift 1 ;;
+        --enable-input-lock)
+                            ENABLE_INPUT_LOCK=1
+                            add_var_to_config ENABLE_INPUT_LOCK "$ENABLE_INPUT_LOCK"
+                            shift 1 ;;
                  --daemon)
                             DAEMON="1"
                             QUIET="1"
@@ -563,6 +591,11 @@ process_arguments () {
                             is_var_empty "$2"
                             add_var_to_config DAEMON_FILE_AGE "$DAEMON_FILE_AGE"
                             shift 2 ;;
+                --email|-e)
+                            is_var_empty "$2"
+                            EMAIL="$2"
+                            add_var_to_config EMAIL "$EMAIL"
+                            shift 2 ;;
           --awskeypair|-P)
                 AWS_KEYPAIR="$2"
                 add_var_to_config AWS_KEYPAIR "$AWS_KEYPAIR"
@@ -1435,13 +1468,9 @@ random_delay () {
     fi
 
     NUMBER=$RANDOM
+
     let "NUMBER %= $ARGS"
-    if [ "$ARCH" == "SunOS" ]
-    then
     sleep "$NUMBER"
-    else
-        sleep "0.$NUMBER"
-    fi
 }
 
 escape_item () {
@@ -1709,7 +1738,7 @@ remove_processed_items_from_input_file () {
 
 get_all_items () {
 
-    if [ "$DAEMON" == "1" ]
+    if [ "$DAEMON" == "1" ] && [ "$INOTIFY" = "0" ]  && [ "$ENABLE_INPUT_LOCK" = "1" ]
     then
         GLOBAL_COUNTER=1
         get_input_lock
@@ -1900,7 +1929,24 @@ elapsed () {
     MINS="$(expr $(expr $REMAINDER - $SECS) / 60)"
 
     RES=$(printf "Total processing time (hh:mm:ss): %02d:%02d:%02d" $HOURS $MINS $SECS)
-    log DSPLY "$RES"
+    echo "$RES"
+}
+
+mail_on_error () {
+
+    ITEM="$1"
+    LOGFILE="$2"
+
+    if [ "$MAIL_ON_ERROR" = "1" ]
+    then
+        cat "$LOGFILE" | mail -s "$HOSTNAME - PPSS: procesing failed for item." "$EMAIL" 
+        if [ "$?" = "0" ]
+        then
+            log DEBUG "Error mail sent."
+        else
+            log ERROR "Sending of error email failed."
+        fi
+    fi
 }
 
 commando () {
@@ -2073,6 +2119,7 @@ commando () {
     # Some error logging. Success or fail.
     if [ ! "$ERROR" == "0" ] 
     then
+       mail_on_error "$ITEM" "$ITEM_LOG_FILE"
        echo -e "Status:\t\tFAILURE" >> "$ITEM_LOG_FILE"
     else
        echo -e "Status:\t\tSUCCESS" >> "$ITEM_LOG_FILE"
@@ -2182,7 +2229,7 @@ run_command () {
             INPUT="$REGISTER"
         fi
 
-        log DEBUG "Now processing $INPUT"
+        log INFO "Now processing $INPUT"
 
         if [ ! -d "$INPUT" ] && [ ! -z "$INPUT" ]
         then
@@ -2217,6 +2264,31 @@ display_jobs_remaining () {
     fi
 }
 
+show_eta () {
+    
+    CURRENT_PROCESSED=$((GLOBAL_COUNTER-MAX_NO_OF_RUNNING_JOBS))
+    TOTAL="$SIZE_OF_INPUT"
+    START_TIME=$START_PPSS
+    NOW=`get_time_in_seconds`
+
+    if [ "$CURRENT_PROCESSED" -le "0" ]
+    then 
+        return 0
+    else
+        TIME_PER_ITEM=$(( (NOW-START_TIME)/CURRENT_PROCESSED ))
+        TOTAL_TIME=$(( ($TIME_PER_ITEM * SIZE_OF_INPUT) + $TIME_PER_ITEM ))
+        TOTAL_TIME_IN_SECONDS=$((START_TIME+TOTAL_TIME))
+        if [ "$ARCH" = "Darwin" ]
+        then
+            DATE=`date -r $TOTAL_TIME_IN_SECONDS`
+        else
+            DATE=`date -d @$TOTAL_TIME_IN_SECONDS`
+        fi
+        echo
+        log DSPLY "ETA: $DATE"
+    fi
+}
+
 display_progress () {
 
     if [ "$DAEMON" = "0" ]
@@ -2228,6 +2300,7 @@ display_progress () {
             if [ "$QUIET" == "0" ]
             then
                 log PRCNT "Currently $PERCENT percent complete. Processed $GLOBAL_COUNTER of $SIZE_OF_INPUT items." 
+                show_eta
             elif [ "$DAEMON" == "0" ]
             then
                 echo -en "\r$PERCENT%"
@@ -2251,7 +2324,6 @@ terminate_listener () {
 
     if [ ! -z "$SSH_MASTER_PID" ]
     then
-        log DEBUG "SSH master PID is $SSH_MASTER_PID"
         kill "$SSH_MASTER_PID" 
     else
         log DEBUG "SSH master PID is empty."
@@ -2275,6 +2347,16 @@ terminate_listener () {
     then
         echo 
     fi
+
+    if [ ! -z "$EMAIL" ]  
+    then
+        echo "PPSS job finished." | mail -s "$HOSTNAME - PPSS has finished." "$EMAIL"
+        if [ ! "$?" = "0" ]
+        then 
+            log ERROR "Sending os status mail failed."
+        fi
+    fi
+
     cleanup
 }
 
@@ -2422,7 +2504,7 @@ listen_for_job () {
 
     while read event <& 42
     do
-        log DEBUG "Current active workers is $ACTIVE_WORKERS"
+        log INFO "Current active workers is $ACTIVE_WORKERS"
 
         if [ "$event" = "$START_KEY" ]
         then
@@ -2455,7 +2537,7 @@ listen_for_job () {
             infanticide
             break
         else
-            log DEBUG "Event $event is an item!"
+            log DEBUG "Event is an item."
             stack_push "$event"
             run_command
         fi

ppss-test.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 DEBUG="$1"
-VERSION="2.81"
+VERSION="2.82"
 TMP_DIR="/tmp/ppss"
 PPSS=./ppss
 PPSS_DIR=ppss_dir