Revision 29044
Added by Eri Katsari over 10 years ago
| modules/dnet-openaire-stats/trunk/src/main/java/eu/dnetlib/data/mapreduce/hbase/statsExport/drivers/SqoopDriver.java | ||
|---|---|---|
| 12 | 12 |
import org.apache.hadoop.fs.LocatedFileStatus; |
| 13 | 13 |
import org.apache.hadoop.fs.Path; |
| 14 | 14 |
import org.apache.hadoop.fs.RemoteIterator; |
| 15 |
import org.apache.log4j.BasicConfigurator; |
|
| 15 | 16 |
import org.apache.log4j.Logger; |
| 16 | 17 |
import org.apache.sqoop.Sqoop; |
| 17 | 18 |
|
| ... | ... | |
| 67 | 68 |
|
| 68 | 69 |
"--verbose", "--username", dbUser, "--password", dbPass, "--driver", "org.postgresql.Driver", "--batch", "--mapreduce-job-name", "Sqoop Stats Import", "-m", sqoopReducersCount }; |
| 69 | 70 |
|
| 70 |
|
|
| 71 |
int ret = Sqoop.runTool(str,conf); |
|
| 72 |
if (ret != 0) {
|
|
| 73 |
log.error("Could not run Sqoop Tool " + Integer.toString(ret));
|
|
| 74 |
throw new RuntimeException("Could not run Sqoop Tool " + Integer.toString(ret) );
|
|
| 75 |
} |
|
| 76 |
|
|
| 71 |
int ret = Sqoop.runTool(str); |
|
| 72 |
if (ret != 0) {
|
|
| 73 |
log.error("Could not run Sqoop Tool " + Integer.toString(ret));
|
|
| 74 |
throw new RuntimeException("Could not run Sqoop Tool " + Integer.toString(ret));
|
|
| 75 |
} |
|
| 76 |
|
|
| 77 | 77 |
cleanUp(table.getKey()); |
| 78 | 78 |
} |
| 79 | 79 |
} |
| ... | ... | |
| 94 | 94 |
for (Entry<Object, Object> e : tableMappings.entrySet()) {
|
| 95 | 95 |
String name = (String) e.getKey(); |
| 96 | 96 |
for (String filename : fileNames) {
|
| 97 |
String str=filename.substring(filename.lastIndexOf('\\')+1);
|
|
| 98 |
|
|
| 99 |
String split[] = str.split("-");
|
|
| 97 |
String str = filename.substring(filename.lastIndexOf('\\') + 1);
|
|
| 100 | 98 |
|
| 99 |
String split[] = str.split("-");
|
|
| 100 |
|
|
| 101 | 101 |
split[0] = split[0].replaceAll(".*/", "");
|
| 102 |
log.info(split[0]);
|
|
| 102 |
log.info(split[0]);
|
|
| 103 | 103 |
if (split[0].equals(name)) {
|
| 104 | 104 |
|
| 105 | 105 |
tables.put((String) e.getValue(), filename); |
| ... | ... | |
| 125 | 125 |
} |
| 126 | 126 |
|
| 127 | 127 |
public ArrayList<String> listHdfsDir() throws Exception {
|
| 128 |
if (conf == null) {
|
|
| 129 |
conf = new Configuration(); |
|
| 130 |
} |
|
| 128 | 131 |
hdfs = FileSystem.get(conf); |
| 129 | 132 |
|
| 130 | 133 |
RemoteIterator<LocatedFileStatus> Files; |
| ... | ... | |
| 189 | 192 |
tableMappings.load(file); |
| 190 | 193 |
file.close(); |
| 191 | 194 |
if (tableMappings == null || tableMappings.isEmpty()) {
|
| 192 |
throw new Exception("Could not load Table Mappings in sqoop init job");
|
|
| 195 |
throw new Exception("Could not load Table Mconfappings in sqoop init job");
|
|
| 193 | 196 |
} |
| 194 | 197 |
} |
| 195 | 198 |
|
| ... | ... | |
| 217 | 220 |
} |
| 218 | 221 |
} |
| 219 | 222 |
|
| 223 |
public static void main(String[] args) {
|
|
| 224 |
SqoopDriver sqoopDriver = new SqoopDriver(); |
|
| 225 |
|
|
| 226 |
BasicConfigurator.configure(); |
|
| 227 |
|
|
| 228 |
sqoopDriver.setDelim("!");
|
|
| 229 |
sqoopDriver.setOutputPath("/tmp/test_stats/");
|
|
| 230 |
|
|
| 231 |
sqoopDriver.setConnectionUrl("jdbc:postgresql://duffy.di.uoa.gr:5432/test_stats");
|
|
| 232 |
sqoopDriver.setDbUser("sqoop");
|
|
| 233 |
sqoopDriver.setDbPass("sqoop");
|
|
| 234 |
sqoopDriver.setReducersCount("1");
|
|
| 235 |
sqoopDriver.setRecsPerStatement("1000");
|
|
| 236 |
sqoopDriver.setStatementPerTrans("1000");
|
|
| 237 |
|
|
| 238 |
sqoopDriver.setUseHdfsStore(true); |
|
| 239 |
|
|
| 240 |
// HashMap<String, String> tables = new HashMap<String, String>(); |
|
| 241 |
|
|
| 242 |
// tables.put("datasource", "/tmp/test_stats/datasource-r-00000");
|
|
| 243 |
|
|
| 244 |
// throw new Exception(datasource + " " + e + |
|
| 245 |
// sqoopDriver.getConnectionUrl() + sqoopDriver.getDbPass() + |
|
| 246 |
// sqoopDriver.getDbUser() ); |
|
| 247 |
|
|
| 248 |
try {
|
|
| 249 |
sqoopDriver.initSqoopJob(); |
|
| 250 |
// sqoopDriver.run(tables); |
|
| 251 |
} catch (Exception e) {
|
|
| 252 |
System.out.print("ERROR " + e.toString());
|
|
| 253 |
e.printStackTrace(); |
|
| 254 |
} |
|
| 255 |
|
|
| 256 |
} |
|
| 257 |
|
|
| 220 | 258 |
public String getConnectionUrl() {
|
| 221 | 259 |
return connectionUrl; |
| 222 | 260 |
} |
| modules/dnet-openaire-stats/trunk/src/main/resources/eu/dnetlib/data/mapreduce/hbase/statsExport/log4j.properties | ||
|---|---|---|
| 15 | 15 |
log4j.appender.stdout.layout=org.apache.log4j.PatternLayout |
| 16 | 16 |
log4j.appender.stdout.layout.ConversionPattern=%d{yyyy-MM-dd HH:mm:ss} %-5p %c{1}:%L - %m%n
|
| 17 | 17 |
|
| 18 |
#log4j.rootLogger = WARN, R |
|
| 19 |
|
|
| 20 |
#log4j.logger.gr.uoa = DEBUG |
|
| 21 |
#log4j.logger.org.springframework = DEBUG, S |
|
| 22 |
#log4j.additivity.org.springframework = false |
|
| 23 |
# |
|
| 24 |
#log4j.logger.com.opensymphony.xwork2.ognl.OgnlValueStack = FATAL |
|
| 25 |
#log4j.logger.com.opensymphony.xwork2.ObjectFactory = FATAL |
|
| 26 |
# |
|
| 27 |
#log4j.appender.R=org.apache.log4j.RollingFileAppender |
|
| 28 |
#log4j.appender.R.File=/tmp/log/validator.log |
|
| 29 |
#log4j.appender.R.MaxFileSize=10MB |
|
| 30 |
#log4j.appender.R.MaxBackupIndex=10 |
|
| 31 |
#log4j.appender.R.layout=org.apache.log4j.PatternLayout |
|
| 32 |
#log4j.appender.R.layout.ConversionPattern= %d %p %t [%c] - %m%n |
|
| 33 |
# |
|
| 34 |
#log4j.appender.S=org.apache.log4j.RollingFileAppender |
|
| 35 |
#log4j.appender.S.File=/tmp/log/spring-validator.log |
|
| 36 |
#log4j.appender.S.MaxFileSize=10MB |
|
| 37 |
#log4j.appender.S.MaxBackupIndex=10 |
|
| 38 |
#log4j.appender.S.layout=org.apache.log4j.PatternLayout |
|
| 39 |
#log4j.appender.S.layout.ConversionPattern= %d %p %t [%c] - %m%n |
|
| modules/dnet-openaire-stats/trunk/pom.xml | ||
|---|---|---|
| 12 | 12 |
<version>0.0.1-SNAPSHOT</version> |
| 13 | 13 |
<build> |
| 14 | 14 |
<plugins> |
| 15 |
|
|
| 15 |
|
|
| 16 |
<plugin> |
|
| 17 |
<groupId>org.codehaus.mojo</groupId> |
|
| 18 |
<artifactId>exec-maven-plugin</artifactId> |
|
| 19 |
<version>1.2.1</version> |
|
| 20 |
<configuration> |
|
| 21 |
<mainClass>eu.dnetlib.data.mapreduce.hbase.statsExport.drivers.SqoopDriver</mainClass> |
|
| 22 |
</configuration> |
|
| 23 |
</plugin> |
|
| 24 |
<plugin> |
|
| 25 |
<groupId>org.apache.maven.plugins</groupId> |
|
| 26 |
<artifactId>maven-shade-plugin</artifactId> |
|
| 27 |
<version>2.1</version> |
|
| 28 |
<executions> |
|
| 29 |
<execution> |
|
| 30 |
<phase>package</phase> |
|
| 31 |
<goals> |
|
| 32 |
<goal>shade</goal> |
|
| 33 |
</goals> |
|
| 34 |
<configuration> |
|
| 35 |
<transformers> |
|
| 36 |
<transformer |
|
| 37 |
implementation="org.apache.maven.plugins.shade.resource.ManifestResourceTransformer"> |
|
| 38 |
</transformer> |
|
| 39 |
</transformers> |
|
| 40 |
</configuration> |
|
| 41 |
</execution> |
|
| 42 |
</executions> </plugin> |
|
| 16 | 43 |
</plugins> |
| 44 |
|
|
| 17 | 45 |
</build> |
| 18 |
<dependencies>
|
|
| 46 |
<dependencies> |
|
| 19 | 47 |
|
| 20 |
<dependency> |
|
| 21 |
<groupId>jdk.tools</groupId> |
|
| 22 |
<artifactId>jdk.tools</artifactId> |
|
| 23 |
<version>1.7.0_05</version> |
|
| 24 |
<scope>system</scope> |
|
| 25 |
<systemPath>${JAVA_HOME}/lib/tools.jar</systemPath>
|
|
| 26 |
</dependency> |
|
| 27 |
|
|
| 28 |
|
|
| 29 | 48 |
<dependency> |
| 49 |
<groupId>jdk.tools</groupId> |
|
| 50 |
<artifactId>jdk.tools</artifactId> |
|
| 51 |
<version>1.7.0_05</version> |
|
| 52 |
<scope>system</scope> |
|
| 53 |
<systemPath>${JAVA_HOME}/lib/tools.jar</systemPath>
|
|
| 54 |
</dependency> |
|
| 55 |
<dependency> |
|
| 56 |
<groupId>org.apache.logging.log4j</groupId> |
|
| 57 |
<artifactId>log4j-api</artifactId> |
|
| 58 |
<version>2.0-rc2</version> |
|
| 59 |
</dependency> |
|
| 60 |
<dependency> |
|
| 61 |
<groupId>org.apache.logging.log4j</groupId> |
|
| 62 |
<artifactId>log4j-core</artifactId> |
|
| 63 |
<version>2.0-rc2</version> |
|
| 64 |
</dependency> |
|
| 65 |
<dependency> |
|
| 66 |
<groupId>log4j</groupId> |
|
| 67 |
<artifactId>log4j</artifactId> |
|
| 68 |
<version>1.2.17</version> |
|
| 69 |
</dependency> |
|
| 70 |
<dependency> |
|
| 30 | 71 |
<groupId>org.slf4j</groupId> |
| 31 | 72 |
<artifactId>slf4j-api</artifactId> |
| 32 | 73 |
<version>1.7.6</version> |
Also available in: Unified diff
m