version 1.2.1
The problem arises *only* when the first query is not finished and you fire second query.
When you submit a sql query, the query is not finished, if you submit another query,
there will be a exception
09/03/03 15:20:59 INFO mapred.JobClient: Task Id : attempt_200903031500_0006_r_000000_1, Status : FAILED
java.io.IOException: File: hdfs://hadoop2.xunjienet.com:9000/user/hadoop/cloudbase/lib/commons-logging-1.0.4.jar has changed on HDFS since job started
at org.apache.hadoop.filecache.DistributedCache.ifExistsAndFresh(DistributedCache.java:450)
at org.apache.hadoop.filecache.DistributedCache.localizeCache(DistributedCache.java:341)
at org.apache.hadoop.filecache.DistributedCache.getLocalCache(DistributedCache.java:203)
at org.apache.hadoop.mapred.TaskRunner.run(TaskRunner.java:178)
The file hdfs://hadoop2.xunjienet.com:9000/user/hadoop/cloudbase/lib/commons-logging-1.0.4.jar timestamp is always change, but when I use 1.1, the jar file timestamp is not change always.
Working on this... will submit a patch soon. Meanwhile one can use this workaround-
if the goal is to submit multiple queries so that they run *one after the another*, then append semi-colon at the end of each query. This will make Squirrel to send subsequent queries to the CloudBase only when the previous query is finished, for example-
SELECT * FROM test_table1;
SELECT * FROM test_table2;
Fixed in version 1.3