summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--IP-learning-graph/src/main/java/cn/ac/iie/service/ingestion/ReadClickhouseData.java24
-rw-r--r--ip-learning-spark/src/main/scala/cn/ac/iie/dao/BaseClickhouseData.scala4
-rw-r--r--ip-learning-spark/src/main/scala/cn/ac/iie/main/IpRecommendApplication.scala2
-rw-r--r--ip-learning-spark/src/main/scala/cn/ac/iie/main/SubscriberRecommendApplication.scala2
-rw-r--r--ip-learning-spark/src/main/scala/cn/ac/iie/service/transform/MergeDataFrame.scala10
-rw-r--r--ip-learning-spark/src/main/scala/cn/ac/iie/service/update/UpdateDocument.scala36
-rw-r--r--ip-learning-spark/src/test/scala/cn/ac/iie/dao/BaseClickhouseDataTest.scala4
7 files changed, 27 insertions, 55 deletions
diff --git a/IP-learning-graph/src/main/java/cn/ac/iie/service/ingestion/ReadClickhouseData.java b/IP-learning-graph/src/main/java/cn/ac/iie/service/ingestion/ReadClickhouseData.java
index 39b0439..16b79a6 100644
--- a/IP-learning-graph/src/main/java/cn/ac/iie/service/ingestion/ReadClickhouseData.java
+++ b/IP-learning-graph/src/main/java/cn/ac/iie/service/ingestion/ReadClickhouseData.java
@@ -140,11 +140,11 @@ public class ReadClickhouseData {
try {
String vFqdn = resultSet.getString("FQDN");
if (isDomain(vFqdn)) {
- String vIp = resultSet.getString("common_server_ip");
+ String vIp = resultSet.getString("server_ip");
long firstFoundTime = resultSet.getLong("FIRST_FOUND_TIME");
long lastFoundTime = resultSet.getLong("LAST_FOUND_TIME");
long countTotal = resultSet.getLong("COUNT_TOTAL");
- String schemaType = resultSet.getString("schema_type");
+ String schemaType = resultSet.getString("decoded_as");
String[] distCipRecents = (String[]) resultSet.getArray("DIST_CIP_RECENT").getArray();
long[] clientIpTs = new long[distCipRecents.length];
for (int i = 0; i < clientIpTs.length; i++) {
@@ -174,12 +174,12 @@ public class ReadClickhouseData {
try {
String vFqdn = resultSet.getString("FQDN");
if (isDomain(vFqdn)) {
- String vIp = resultSet.getString("common_client_ip");
+ String vIp = resultSet.getString("client_ip");
String key = vIp + "-" + vFqdn;
long firstFoundTime = resultSet.getLong("FIRST_FOUND_TIME");
long lastFoundTime = resultSet.getLong("LAST_FOUND_TIME");
long countTotal = resultSet.getLong("COUNT_TOTAL");
- String schemaType = resultSet.getString("schema_type");
+ String schemaType = resultSet.getString("decoded_as");
newDoc = new BaseEdgeDocument();
newDoc.setKey(key);
@@ -257,29 +257,29 @@ public class ReadClickhouseData {
public static String getVertexFqdnSql() {
String where = "recv_time >= " + minTime + " AND recv_time < " + maxTime;
- String sslSql = "SELECT ssl_sni AS FQDN,MAX( recv_time ) AS LAST_FOUND_TIME,MIN( recv_time ) AS FIRST_FOUND_TIME FROM tsg_galaxy_v3.session_record WHERE " + where + " and common_schema_type = 'SSL' GROUP BY ssl_sni";
- String httpSql = "SELECT http_host AS FQDN,MAX( recv_time ) AS LAST_FOUND_TIME,MIN( recv_time ) AS FIRST_FOUND_TIME FROM tsg_galaxy_v3.session_record WHERE " + where + " and common_schema_type = 'HTTP' GROUP BY http_host";
+ String sslSql = "SELECT ssl_sni AS FQDN,MAX( recv_time ) AS LAST_FOUND_TIME,MIN( recv_time ) AS FIRST_FOUND_TIME FROM tsg_galaxy_v3.session_record WHERE " + where + " and decoded_as = 'SSL' GROUP BY ssl_sni";
+ String httpSql = "SELECT http_host AS FQDN,MAX( recv_time ) AS LAST_FOUND_TIME,MIN( recv_time ) AS FIRST_FOUND_TIME FROM tsg_galaxy_v3.session_record WHERE " + where + " and decoded_as = 'HTTP' GROUP BY http_host";
return "SELECT FQDN,MAX( LAST_FOUND_TIME ) AS LAST_FOUND_TIME,MIN( FIRST_FOUND_TIME ) AS FIRST_FOUND_TIME FROM ((" + sslSql + ") UNION ALL (" + httpSql + ")) GROUP BY FQDN HAVING FQDN != ''";
}
public static String getVertexIpSql() {
String where = " recv_time >= " + minTime + " AND recv_time < " + maxTime;
- String clientIpSql = "SELECT common_client_ip AS IP, MIN(recv_time) AS FIRST_FOUND_TIME,MAX(recv_time) AS LAST_FOUND_TIME,count(*) as SESSION_COUNT,sum(common_c2s_byte_num+common_s2c_byte_num) as BYTES_SUM,groupUniqArray(2)(common_link_info_c2s) as common_link_info,'client' as ip_type FROM tsg_galaxy_v3.session_record where " + where + " group by IP";
- String serverIpSql = "SELECT common_server_ip AS IP, MIN(recv_time) AS FIRST_FOUND_TIME,MAX(recv_time) AS LAST_FOUND_TIME,count(*) as SESSION_COUNT,sum(common_c2s_byte_num+common_s2c_byte_num) as BYTES_SUM,groupUniqArray(2)(common_link_info_s2c) as common_link_info,'server' as ip_type FROM tsg_galaxy_v3.session_record where " + where + " group by IP";
+ String clientIpSql = "SELECT client_ip AS IP, MIN(recv_time) AS FIRST_FOUND_TIME,MAX(recv_time) AS LAST_FOUND_TIME,count(*) as SESSION_COUNT,sum(sent_bytes+received_bytes) as BYTES_SUM,groupUniqArray(2)(common_link_info_c2s) as common_link_info,'client' as ip_type FROM tsg_galaxy_v3.session_record where " + where + " group by IP";
+ String serverIpSql = "SELECT server_ip AS IP, MIN(recv_time) AS FIRST_FOUND_TIME,MAX(recv_time) AS LAST_FOUND_TIME,count(*) as SESSION_COUNT,sum(sent_bytes+received_bytes) as BYTES_SUM,groupUniqArray(2)(common_link_info_s2c) as common_link_info,'server' as ip_type FROM tsg_galaxy_v3.session_record where " + where + " group by IP";
return "SELECT * FROM((" + clientIpSql + ") UNION ALL (" + serverIpSql + "))";
}
public static String getRelationshipFqdnAddressIpSql() {
String where = " recv_time >= " + minTime + " AND recv_time < " + maxTime;
- String sslSql = "SELECT ssl_sni AS FQDN,common_server_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,groupUniqArray("+DISTINCT_CLIENT_IP_NUM+")(common_client_ip) AS DIST_CIP_RECENT,'TLS' AS schema_type FROM tsg_galaxy_v3.session_record WHERE " + where + " and common_schema_type = 'SSL' GROUP BY ssl_sni,common_server_ip";
- String httpSql = "SELECT http_host AS FQDN,common_server_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,groupUniqArray("+DISTINCT_CLIENT_IP_NUM+")(common_client_ip) AS DIST_CIP_RECENT,'HTTP' AS schema_type FROM tsg_galaxy_v3.session_record WHERE " + where + " and common_schema_type = 'HTTP' GROUP BY http_host,common_server_ip";
+ String sslSql = "SELECT ssl_sni AS FQDN,server_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,groupUniqArray("+DISTINCT_CLIENT_IP_NUM+")(client_ip) AS DIST_CIP_RECENT,'TLS' AS decoded_as FROM tsg_galaxy_v3.session_record WHERE " + where + " and decoded_as = 'SSL' GROUP BY ssl_sni,server_ip";
+ String httpSql = "SELECT http_host AS FQDN,server_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,groupUniqArray("+DISTINCT_CLIENT_IP_NUM+")(client_ip) AS DIST_CIP_RECENT,'HTTP' AS decoded_as FROM tsg_galaxy_v3.session_record WHERE " + where + " and decoded_as = 'HTTP' GROUP BY http_host,server_ip";
return "SELECT * FROM ((" + sslSql + ") UNION ALL (" + httpSql + "))WHERE FQDN != ''";
}
public static String getRelationshipIpVisitFqdnSql() {
String where = " recv_time >= " + minTime + " AND recv_time < " + maxTime;
- String httpSql = "SELECT http_host AS FQDN,common_client_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,'HTTP' AS schema_type FROM tsg_galaxy_v3.session_record WHERE " + where + " and common_schema_type = 'HTTP' GROUP BY http_host,common_client_ip";
- String sslSql = "SELECT ssl_sni AS FQDN,common_client_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,'TLS' AS schema_type FROM tsg_galaxy_v3.session_record WHERE common_schema_type = 'SSL' GROUP BY ssl_sni,common_client_ip";
+ String httpSql = "SELECT http_host AS FQDN,client_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,'HTTP' AS decoded_as FROM tsg_galaxy_v3.session_record WHERE " + where + " and decoded_as = 'HTTP' GROUP BY http_host,client_ip";
+ String sslSql = "SELECT ssl_sni AS FQDN,client_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,'TLS' AS decoded_as FROM tsg_galaxy_v3.session_record WHERE decoded_as = 'SSL' GROUP BY ssl_sni,client_ip";
return "SELECT * FROM ((" + sslSql + ") UNION ALL (" + httpSql + "))WHERE FQDN != ''";
}
diff --git a/ip-learning-spark/src/main/scala/cn/ac/iie/dao/BaseClickhouseData.scala b/ip-learning-spark/src/main/scala/cn/ac/iie/dao/BaseClickhouseData.scala
index a6c691d..cfe8b5e 100644
--- a/ip-learning-spark/src/main/scala/cn/ac/iie/dao/BaseClickhouseData.scala
+++ b/ip-learning-spark/src/main/scala/cn/ac/iie/dao/BaseClickhouseData.scala
@@ -100,12 +100,12 @@ object BaseClickhouseData {
s"""
|(SELECT * FROM
|((SELECT ssl_sni AS FQDN,server_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,
- |toString(groupUniqArray(${ApplicationConfig.DISTINCT_CLIENT_IP_NUM})(client_ip)) AS DIST_CIP_RECENT,'TLS' AS schema_type,vsys_id AS VSYS_ID
+ |toString(groupUniqArray(${ApplicationConfig.DISTINCT_CLIENT_IP_NUM})(client_ip)) AS DIST_CIP_RECENT,'TLS' AS decoded_as,vsys_id AS VSYS_ID
|FROM ${ApplicationConfig.SPARK_READ_CLICKHOUSE_SESSION_TABLE}
|WHERE $where and decoded_as = 'SSL' GROUP BY ssl_sni,server_ip,vsys_id)
|UNION ALL
|(SELECT http_host AS FQDN,server_ip,MAX(recv_time) AS LAST_FOUND_TIME,MIN(recv_time) AS FIRST_FOUND_TIME,COUNT(*) AS COUNT_TOTAL,
- |toString(groupUniqArray(${ApplicationConfig.DISTINCT_CLIENT_IP_NUM})(client_ip)) AS DIST_CIP_RECENT,'HTTP' AS schema_type,vsys_id AS VSYS_ID
+ |toString(groupUniqArray(${ApplicationConfig.DISTINCT_CLIENT_IP_NUM})(client_ip)) AS DIST_CIP_RECENT,'HTTP' AS decoded_as,vsys_id AS VSYS_ID
|FROM ${ApplicationConfig.SPARK_READ_CLICKHOUSE_SESSION_TABLE}
|WHERE $where and decoded_as = 'HTTP' GROUP BY http_host,server_ip,vsys_id))
|WHERE FQDN != '') as dbtable
diff --git a/ip-learning-spark/src/main/scala/cn/ac/iie/main/IpRecommendApplication.scala b/ip-learning-spark/src/main/scala/cn/ac/iie/main/IpRecommendApplication.scala
index 99b188d..7fa6d62 100644
--- a/ip-learning-spark/src/main/scala/cn/ac/iie/main/IpRecommendApplication.scala
+++ b/ip-learning-spark/src/main/scala/cn/ac/iie/main/IpRecommendApplication.scala
@@ -1,7 +1,7 @@
package cn.ac.iie.main
import cn.ac.iie.service.update.UpdateDocument
-@deprecated
+
object IpRecommendApplication {
def main(args: Array[String]): Unit = {
diff --git a/ip-learning-spark/src/main/scala/cn/ac/iie/main/SubscriberRecommendApplication.scala b/ip-learning-spark/src/main/scala/cn/ac/iie/main/SubscriberRecommendApplication.scala
index 3ee06b2..04cbec4 100644
--- a/ip-learning-spark/src/main/scala/cn/ac/iie/main/SubscriberRecommendApplication.scala
+++ b/ip-learning-spark/src/main/scala/cn/ac/iie/main/SubscriberRecommendApplication.scala
@@ -1,7 +1,7 @@
package cn.ac.iie.main
import cn.ac.iie.service.update.UpdateDocument
-@deprecated
+
object SubscriberRecommendApplication {
def main(args: Array[String]): Unit = {
diff --git a/ip-learning-spark/src/main/scala/cn/ac/iie/service/transform/MergeDataFrame.scala b/ip-learning-spark/src/main/scala/cn/ac/iie/service/transform/MergeDataFrame.scala
index 352a650..20ca1e8 100644
--- a/ip-learning-spark/src/main/scala/cn/ac/iie/service/transform/MergeDataFrame.scala
+++ b/ip-learning-spark/src/main/scala/cn/ac/iie/service/transform/MergeDataFrame.scala
@@ -39,7 +39,7 @@ object MergeDataFrame {
)
val ipRddRow = frame.rdd.map(row => {
- val vsysId = row.getAs[Long]("VSYS_ID")
+ val vsysId = row.getAs[Integer]("VSYS_ID").toLong
val ip = row.getAs[String]("IP")
( ip + "-" + vsysId, row)
}) /*.partitionBy(new CustomPartitioner(ApplicationConfig.SPARK_SQL_SHUFFLE_PARTITIONS))*/
@@ -51,18 +51,18 @@ object MergeDataFrame {
def mergeRelationFqdnLocateIp(): RDD[(String, (Option[BaseEdgeDocument], Row))] = {
val frame = BaseClickhouseData.getRelationFqdnLocateIpDf
.repartition().filter(row => isDomain(row.getAs[String]("FQDN")))
- .groupBy("FQDN", "common_server_ip", "VSYS_ID")
+ .groupBy("FQDN", "server_ip", "VSYS_ID")
.agg(
min("FIRST_FOUND_TIME").alias("FIRST_FOUND_TIME"),
max("LAST_FOUND_TIME").alias("LAST_FOUND_TIME"),
collect_list("COUNT_TOTAL").alias("COUNT_TOTAL_LIST"),
- collect_list("schema_type").alias("schema_type_list"),
+ collect_list("decoded_as").alias("decoded_as_list"),
collect_set("DIST_CIP_RECENT").alias("DIST_CIP_RECENT")
)
val fqdnLocIpRddRow = frame.rdd.map(row => {
val fqdn = row.getAs[String]("FQDN")
- val serverIp = row.getAs[String]("common_server_ip")
- val vsysId = row.getAs[Long]("VSYS_ID")
+ val serverIp = row.getAs[String]("server_ip")
+ val vsysId = row.getAs[Integer]("VSYS_ID").toLong
val key = fqdn.concat("-" + serverIp + "-" + vsysId)
(key, row)
}) /*.partitionBy(new CustomPartitioner(ApplicationConfig.SPARK_SQL_SHUFFLE_PARTITIONS))*/
diff --git a/ip-learning-spark/src/main/scala/cn/ac/iie/service/update/UpdateDocument.scala b/ip-learning-spark/src/main/scala/cn/ac/iie/service/update/UpdateDocument.scala
index 2bd7732..96b3dcd 100644
--- a/ip-learning-spark/src/main/scala/cn/ac/iie/service/update/UpdateDocument.scala
+++ b/ip-learning-spark/src/main/scala/cn/ac/iie/service/update/UpdateDocument.scala
@@ -192,34 +192,6 @@ object UpdateDocument {
subidDoc
}
- private def getVertexFqdnRow(joinRow: (String, (Option[BaseDocument], Row))): BaseDocument = {
- val fqdnDocOpt = joinRow._2._1
- var fqdnDoc = fqdnDocOpt match {
- case Some(doc) => doc
- case None => null
- }
- val fqdnRow: Row = joinRow._2._2
- if (fqdnRow != null) {
- val fqdn = fqdnRow.getAs[String]("FQDN")
- val lastFoundTime = fqdnRow.getAs[Long]("LAST_FOUND_TIME")
- val firstFoundTime = fqdnRow.getAs[Long]("FIRST_FOUND_TIME")
- val vsysId = fqdnRow.getAs[Long]("VSYS_ID")
-
- if (fqdnDoc != null) {
- updateMaxAttribute(fqdnDoc, lastFoundTime, "LAST_FOUND_TIME")
- fqdnDoc.addAttribute("VSYS_ID", vsysId)
- } else {
- fqdnDoc = new BaseDocument
- fqdnDoc.setKey(fqdn + "-" + vsysId)
- fqdnDoc.addAttribute("FQDN_NAME", fqdn)
- fqdnDoc.addAttribute("FIRST_FOUND_TIME", firstFoundTime)
- fqdnDoc.addAttribute("LAST_FOUND_TIME", lastFoundTime)
- fqdnDoc.addAttribute("VSYS_ID", vsysId)
- }
- }
- fqdnDoc
- }
-
private def getVertexIpRow(joinRow: (String, (Option[BaseDocument], Row))): BaseDocument = {
val ipDocOpt = joinRow._2._1
var ipDoc = ipDocOpt match {
@@ -236,7 +208,7 @@ object UpdateDocument {
val ipTypeList = ipRow.getAs[ofRef[String]]("ip_type_list")
val linkInfo = ipRow.getAs[String]("common_link_info")
val sepAttributeTuple = separateAttributeByIpType(ipTypeList, sessionCountList, bytesSumList)
- val vsysId = ipRow.getAs[Long]("VSYS_ID")
+ val vsysId = ipRow.getAs[Integer]("VSYS_ID").toLong
if (ipDoc != null) {
updateMaxAttribute(ipDoc, lastFoundTime, "LAST_FOUND_TIME")
@@ -276,13 +248,13 @@ object UpdateDocument {
}
if (fqdnLocIpRow != null) {
val fqdn = fqdnLocIpRow.getAs[String]("FQDN")
- val serverIp = fqdnLocIpRow.getAs[String]("common_server_ip")
+ val serverIp = fqdnLocIpRow.getAs[String]("server_ip")
val firstFoundTime = fqdnLocIpRow.getAs[Long]("FIRST_FOUND_TIME")
val lastFoundTime = fqdnLocIpRow.getAs[Long]("LAST_FOUND_TIME")
val countTotalList = fqdnLocIpRow.getAs[ofRef[AnyRef]]("COUNT_TOTAL_LIST")
- val schemaTypeList = fqdnLocIpRow.getAs[ofRef[AnyRef]]("schema_type_list")
+ val schemaTypeList = fqdnLocIpRow.getAs[ofRef[AnyRef]]("decoded_as_list")
val distCipRecent = fqdnLocIpRow.getAs[ofRef[String]]("DIST_CIP_RECENT")
- val vsysId = fqdnLocIpRow.getAs[Long]("VSYS_ID")
+ val vsysId = fqdnLocIpRow.getAs[Integer]("VSYS_ID").toLong
val sepAttritubeMap: Map[String, Long] = separateAttributeByProtocol(schemaTypeList, countTotalList)
val distinctIp: Array[String] = mergeDistinctIp(distCipRecent)
diff --git a/ip-learning-spark/src/test/scala/cn/ac/iie/dao/BaseClickhouseDataTest.scala b/ip-learning-spark/src/test/scala/cn/ac/iie/dao/BaseClickhouseDataTest.scala
index 01804eb..3fcf09f 100644
--- a/ip-learning-spark/src/test/scala/cn/ac/iie/dao/BaseClickhouseDataTest.scala
+++ b/ip-learning-spark/src/test/scala/cn/ac/iie/dao/BaseClickhouseDataTest.scala
@@ -19,7 +19,7 @@ object BaseClickhouseDataTest {
| FROM
| global_temp.dbtable
| WHERE
- | common_schema_type = 'SSL' GROUP BY ssl_sni
+ | decoded_as = 'SSL' GROUP BY ssl_sni
| )
| UNION ALL
| (SELECT
@@ -27,7 +27,7 @@ object BaseClickhouseDataTest {
| FROM
| global_temp.dbtable
| WHERE
- | common_schema_type = 'HTTP' GROUP BY http_host
+ | decoded_as = 'HTTP' GROUP BY http_host
| )
| )
|GROUP BY