Hi, kylin developers. My cube building procedure failing at Step 2
"Redistribute Flat Hive Table" because Kylin always trying to submit a mr job
to the default yarn queue.
I have overrided the mapred.job.queue.name property in both kylin_hive_conf.xml
and kylin.properties but it doesn't work.
kylin.properties
```
kylin.source.hive.beeline-params=-n hive -p hive --hiveconf
mapred.job.queue.name=myQueue -u
"jdbc:hive2://myZk:2181/;serviceDiscoveryMode=zooKeeper;"
```
kylin_hive_conf.xml
```
<property>
<name>mapred.job.queue.name</name>
<value>myQueue</value>
</property>
```
After digging into the source code ,I found that kylin try to get the row count
of hive table before redistributing it. But it dose not override the hive
configuration when using jdbc to connect to hive server. I wonder if it is a
bug or there is some solutions?
```
public BeelineHiveClient(String beelineParams) {
if (StringUtils.isEmpty(beelineParams)) {
throw new IllegalArgumentException("BeelineParames cannot be
empty");
}
String[] splits = StringUtils.split(beelineParams);
String url = null, username = null, password = null;
for (int i = 0; i < splits.length; i++) {
if ("-u".equals(splits[i])) {
url = stripQuotes(splits[i + 1]);
}
if ("-n".equals(splits[i])) {
username = stripQuotes(splits[i + 1]);
}
if ("-p".equals(splits[i])) {
password = stripQuotes(splits[i + 1]);
}
}
this.init(url, username, password);
}
private void init(String url, String username, String password) {
try {
Class.forName("org.apache.hive.jdbc.HiveDriver");
cnct = DriverManager.getConnection(url, username, password);
//use DriverManager.getConnection(url,properites) here to override hive
configuration settings?
stmt = cnct.createStatement();
metaData = cnct.getMetaData();
} catch (SQLException | ClassNotFoundException e) {
throw new RuntimeException(e);
}
}
```
