[jira] [Created] (OOZIE-3135) Configure log4j2 in SqoopMain

["Attila Sasvari (JIRA)" <ji...@apache.org>]

Attila Sasvari created OOZIE-3135:
-------------------------------------

             Summary: Configure log4j2 in SqoopMain
                 Key: OOZIE-3135
                 URL: https://issues.apache.org/jira/browse/OOZIE-3135
             Project: Oozie
          Issue Type: Bug
    Affects Versions: 5.0.0b1
            Reporter: Attila Sasvari


In Hadoop 3, MAPREDUCE-6983 switched to use slfj4 & log4j2 in {{org.apache.hadoop.mapreduce.Job}} (that prints out MR job id-s needed for Oozie). We need to setup log4j accordingly (it is also related to HADOOP-12956).
Without proper configuration in the Sqoop action, we won't be able to get external job id-s (SqoopActionExecutor unit tests and real action would be also affected).
   
[The API for Log4j 2 is not compatible with Log4j 1.x|https://logging.apache.org/log4j/2.x/], but we will need to support both hadoop 2 and hadoop 3 profiles for a while. 

We could use reflection to determine the type of the logger object in {{org.apache.hadoop.mapreduce.Job}} and configure log4j settings based on it, but there might be a better way.

For example we could do something like this:
- add a new method for configuring log4j2:
{code}
    private String setUpSqoopLog4J2(final String rootLogLevel) throws IOException {
        System.out.println("Setting up log4j2");

        final String logFile = getSqoopLogFile();
        final File log4j2Xml = new File(SQOOP_LOG4J2_XML);
        try (Writer writer = new FileWriter(log4j2Xml))
        {
            final String logj2SettingsXml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" +
                    "<Configuration status=\"WARN\">\n" +
                    "    <Appenders>\n" +
                    "        <Console name=\"Console\" target=\"SYSTEM_OUT\">\n" +
                    "            <PatternLayout pattern=\"%d{HH:mm:ss.SSS} [%t] %-5level %logger{36} - %msg%n\"/>\n" +
                    "        </Console>\n" +
                    "        <File name=\"File\" fileName=\"" + logFile + "\">  \n" +
                    "            <PatternLayout pattern=\"%d{HH:mm:ss.SSS} [%t] %-5level %logger{36} - %msg%n\"/>\n" +
                    "        </File> \n" +
                    "    </Appenders>\n" +
                    "    <Loggers>\n" +
                    "        <Root level=\"" + rootLogLevel.toLowerCase() + "\">\n" +
                    "            <AppenderRef ref=\"Console\"/>\n" +
                    "            <AppenderRef ref=\"File\"/>\n" +
                    "        </Root>\n" +
                    "    </Loggers>\n" +
                    "</Configuration>";
            writer.write(logj2SettingsXml);
        }

        System.out.printf("log4j2 configuration file created at %s%n", log4j2Xml.getAbsolutePath());

        final   LoggerContext context = (LoggerContext) LogManager.getContext(false);
        context.setConfigLocation(log4j2Xml.toURI()); // forces log4j2 reconfiguration
        return logFile;
    }
{code}
and call it in the {{run()}} method if the mapreduce client is using slf4j for logging:
{code}
        String logFile;
        // MAPREDUCE-6983 switches to slfj4 & log4j2. Need to setup log4j accordingly
        if (org.apache.hadoop.mapreduce.Job.class.getDeclaredField("LOG").getType().
                isAssignableFrom(org.slf4j.Logger.class)) {
            logFile = setUpSqoopLog4J2(rootLogLevel);
        }
        else {
            logFile = setUpSqoopLog4J(rootLogLevel, logLevel);
        }
{code}




--
This message was sent by Atlassian JIRA
(v6.4.14#64029)