[ 
https://issues.apache.org/jira/browse/KYLIN-3613?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16662015#comment-16662015
 ] 

WangSheng commented on KYLIN-3613:
----------------------------------

Hi, I met the same problem, I found that when deploy coprocessor, kylin get 
wrong hbase conf
{code:java}
    // DeployCoprocessorCLI.java
    private static void initHTableCoprocessor(HTableDescriptor desc) throws 
IOException {
        KylinConfig kylinConfig = KylinConfig.getInstanceFromEnv();
        // here, hbase conf is not corrent
        Configuration hconf = HBaseConnection.getCurrentHBaseConfiguration();
        FileSystem fileSystem = FileSystem.get(hconf);

        String localCoprocessorJar = kylinConfig.getCoprocessorLocalJar();
        Path hdfsCoprocessorJar = 
DeployCoprocessorCLI.uploadCoprocessorJar(localCoprocessorJar, fileSystem, 
null);

        DeployCoprocessorCLI.addCoprocessorOnHTable(desc, hdfsCoprocessorJar);
    }
    
    // HBaseConnection.java
    public static Configuration getCurrentHBaseConfiguration() {
        if (configThreadLocal.get() == null) {
            // when restart kylin, here is correct
            StorageURL storageUrl = 
KylinConfig.getInstanceFromEnv().getStorageUrl();
            configThreadLocal.set(newHBaseConfiguration(storageUrl));
        }
        // after a while, here get wrong config
        return configThreadLocal.get();
    }
{code}
When restart kylin, configThreadLocal.get().get("fs.defaultFS") is hbase 
cluster nameservice, so everything is normal. But after a while, 
configThreadLocal.get().get("fs.defaultFS") became kylin local cluster 
nameservice, so build error, but I not sure why, still debug......


> Kylin with Standalone HBase Cluster (enabled kerberos) could not find the 
> main cluster namespace at  "Create HTable" step
> -------------------------------------------------------------------------------------------------------------------------
>
>                 Key: KYLIN-3613
>                 URL: https://issues.apache.org/jira/browse/KYLIN-3613
>             Project: Kylin
>          Issue Type: Bug
>          Components: Environment 
>    Affects Versions: v2.4.0
>            Reporter: powerinf
>            Priority: Major
>         Attachments: Kylin_standalone_hbase.log
>
>
> I deployed two hadoop cluster(also enabled kerberos ,with cross-realm trust) 
> the main cluster and hbase cluster,Kylin Server can access both clusters 
> using hdfs shell with fully qualifiered path ,can submit MR job to main 
> cluster, and can use hive shell to access data warehouse
> on Kylin Server, the configurations of hadoop and hive points to main 
> cluster,and can access hbase cluster using hbase shell.
> when I build the cube, at "Create HTable" step, it reported the error 
> "java.net.UnknownHostException: ctyunbigdata Set hbase.table.sanity.checks to 
> false 
> at conf or table descriptor if you want to bypass sanity checks",but after I 
> restart Kylin serer , resume it can run normally,Why?
> more detail message on Kylin_standalone_hbase.log



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to