![]() Non-kerberized Setup DbVisualizer (as of version 9.5.5)īelow is an example configuration using DbVisualizer: This updated code is inserted into the final "uber" jar and replaces the copy of .VersionInfo originally found in hadoop-common.jar. + " Trying this.getClass().getClassLoader()") VersionInfoFile + "' in Thread.currentThread().getContextClassLoader()." In my testing, this eliminated the above error in // both DbVisualizer and SQuirrelSQL. My updated code first checks the Thread.currentThread().getContextClassLoader(), but if no InputStream is found // then checks this.getClass().getClassLoader(). ![]() In any event, I have overwritten .VersionInfo in this project to use a more robust approach for loading the properties file. I suspect the difference in behavior between tools boils down to how each chooses to load the "uber" jar. The trouble seems to be caused by the way .VersionInfo attempts to load the properties file using Thread.currentThread().getContextClassLoader(). : Illegal Hadoop Version: Unknown (expected A.B.* format)Īt .(ShimLoader.java:168)Īt .(ShimLoader.java:143)Īt .(ShimLoader.java:129)Īt .(KerberosSaslHelper.java:54)Īt .HiveConnection.createBinaryTransport(HiveConnection.java:414)Īt .HiveConnection.openTransport(HiveConnection.java:191)Īt .HiveConnection.(HiveConnection.java:155)Īt .nnect(HiveDriver.java:105) To do this, the following snippet of code is called: This class attempts to load a file called *-version-info.properties in an effort to determine the current Hadoop version. When connecting to a kerberized cluster, ultimately the class .VersionInfo is loaded. Added support kerberized clusters - 12/19/16.Resolved issue with Zookeeper discovery - 01/27/17.Updated for Apache Hive 3.1.2 - 07/05/20 LATEST.You can download the latest binaries from the releases page: ![]() A full explanation of configuring Kerberos is outside the scope of this project. Configuring tools to work against kerberized clusters is typically more involved and requires an understanding of Kerberos. Please note this jar works with both kerberized and non-kerberized clusters. The goal of this simple maven project is to easily pull the required jars into a single place locally and create an "uber" or "standalone" jar that can be referenced by any JDBC compliant tool. Recent versions of Hive include a "standalone" jar, but even this does not include all required dependencies. I've found that this can be quite cumbersome, typically requiring you to copy jars from the Hadoop cluster to some place locally that can be read by your tool. Sometimes you want to connect your favorite database query or visualization tool to Hive.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |