Add wifi_syslog readme

etlers/ArubaSyslog/src/main/scala/cn/edu/sjtu/omnilab/odh/spark/CleanWIFILog.scala

@@ -0,0 +1,4 @@
+package cn.edu.sjtu.omnilab.odh.spark
+
+
+case class CleanWIFILog(MAC: String, time: Long, code: Int, payload: String) // payload as AP or IP

etlers/ArubaSyslog/src/main/scala/cn/edu/sjtu/omnilab/odh/spark/CleanseWifiLogs.scala

@@ -3,8 +3,6 @@ package cn.edu.sjtu.omnilab.odh.spark
 import cn.edu.sjtu.omnilab.odh.rawfilter.{WIFILogFilter}
 import org.apache.spark.{SparkConf, SparkContext}
 
-case class CleanWIFILog(MAC: String, time: Long, code: Int, payload: String) // payload as AP or IP
-
 
 /**
  * Cleanse raw WIFI syslog into movement data.

porters/wifi_syslog.md

@@ -0,0 +1,29 @@
+# WifiSyslog
+
+This repo contains cleansed data for Aruba Wifi Syslog, at SJTU.
+
+## Data format
+
+Four fields are contained in the cleansed data:
+
+* `User ID`: string, recorded as device MAC address.
+
+* `Timestamp`: long, recorded as the time (in milliseconds) when the logging action is triggered by Aruba system.
+
+* `Action code number`: int, the formated code for different actions triggered by Aruba system.
+
+    - 0: AuthRequest
+    - 1: Deauth
+    - 2: AssocRequest
+    - 3: Disassoc
+    - 4: UserAuth
+    - 5: IPAllocation
+    - 6: IPRecycle
+
+* `Payload`: string, contains three types of payloads for different actions:
+
+    - `Associated AP name`: action 0-3, when the logging action is triggered. AP name format: `BuildName`-`BuildNum`-`Floor`-`APNum`, where the BuildNum may be absent for small buildings.
+
+    - `User account name`: action 4, the identify of JAccount.
+
+    - `Allocated IP address`: action 5-6, the allocated IP address for current user. IP addresses are reused by Aruba systems. The same IP address MIGHT be reallocated to another user when that address is unused for about 30 mins to a hour.

porters/wifi_syslog.sh

@@ -9,6 +9,11 @@ function die () {
     exit 1
 }
 
+# Check permission
+if [ `whoami` != 'omnilab' ]; then
+    die "Need permission of OMNILAB to run. Try user omnilab."
+fi
+
 # Global vars
 BASEDIR=$(dirname $0)/..
 source $BASEDIR/global_config.sh
@@ -17,6 +22,17 @@ source $BASEDIR/global_config.sh
 BINJAR=$BASEDIR/etlers/ArubaSyslog/target/scala-2.10/ArubaSyslog-assembly-1.0.jar
 CLSNAME="cn.edu.sjtu.omnilab.odh.spark.CleanseWifiLogs"
 
+# Check root path for raw data
+if [ ! -d $WIFI_SYSLOG_PATH ]; then
+    die "Cann't find path for raw syslog: $WIFI_SYSLOG_PATH"
+fi
+
+# Check root path for this repo
+if ! hadoop fs -test -d $HDFS_WIFI_SYSLOG; then
+    hadoop fs -mkdir -p $HDFS_WIFI_SYSLOG
+fi
+
+# Clear temporary folder
 TEMPWP=$HDFS_WIFI_SYSLOG/_temp
 hadoop fs -rm -r $TEMPWP
 hadoop fs -mkdir -p $TEMPWP
@@ -32,18 +48,19 @@ for rawfile in `ls $WIFI_SYSLOG_PATH`; do
 
     if [ $year$month$day == $(date -d "yesterday" '+%Y%m%d') ]; then
 
-        if ! hadoop fs -test -e $HDFS_WIFI_SYSLOG/$rfname/_SUCCESS; then
+	if ! hadoop fs -test -e $HDFS_WIFI_SYSLOG/$rfname/_SUCCESS; then
 
-            # Decompress file
-            if ! hadoop fs -test -e $TEMPWP/$rfname; then
-                gunzip -c $WIFI_SYSLOG_PATH/$rawfile | hadoop fs -put - $TEMPWP/$rfname
-            fi
+	    # Decompress file
+	    if ! hadoop fs -test -e $TEMPWP/$rfname; then
+		gunzip -c $WIFI_SYSLOG_PATH/$rawfile | hadoop fs -put - $TEMPWP/$rfname
+	    fi
 
-            # Cleanse wifilog
-            hadoop fs -rm -r $HDFS_WIFI_SYSLOG/$rfname
-            spark-submit2 --class $CLSNAME $BINJAR $TEMPWP/$rfname $HDFS_WIFI_SYSLOG/$rfname
+	    # Cleanse wifilog
+	    hadoop fs -rm -r $HDFS_WIFI_SYSLOG/$rfname
+	    spark-submit2 --class $CLSNAME $BINJAR $TEMPWP/$rfname $HDFS_WIFI_SYSLOG/$rfname
+	    hadoop fs -rm -r $TEMPWP/$rfname
+	fi
 
-        fi
     fi
 
 done