问题
I am installing Cloudera Manager onto an ec2 instance. I only added a single other ec2 instance to the cluster. The installation succeeded, but when the manager initiates the cluster services (step 9 of 21) I get the following error:
[2013-07-12 18:44:35,906]ERROR 63227[main]
com.cloudera.enterprise.dbutil.SqlRunner.open(SqlRunner.java:111)
- Error connecting to db with user 'hive' and jdbcUrl 'jdbc:postgresql://ip-xx-xxx-
xx-x.ec2.internal:7432/hive'
I manually opened port 7432 on the ec2 instance created by cloudera, because it did not appear to be open, I'm not sure if that was a bad idea. The cloudera manager docs claim that the postgres db will be auto created on installation so I don't think that is the problem either.
回答1:
I've been getting this error more and more lately.
Check your Private DNS of the created AMI in EC2 Console and compare it to the JDBC URI from the error. I've found that the Private DNS is incorrect when I get this error, though I have no clue how to get around it.
回答2:
I had the same issue. It turned out that manager instance had different Security Group than the one that was having instances launched by the manager. So, I provided full access to those security group instanced for each other.
It was fixed.
回答3:
It looks like this can be caused by stopping/starting the Cloudera Manager instance, if it comes back up with a new IP address.
I fixed it by doing the following:
- In the Cloudera Manager interface, click the "hive1" service.
- Click Configuration / View and Edit.
- Expand "Service-Wide" and click "Hive Metastore Database".
- Check the "Hive Metastore Database Host" setting - it is probably pointing at an old address you don't have control of anymore.
- Replace that with the Manager instance's current private DNS name, obtained from the EC2 console.
来源：https://stackoverflow.com/questions/17622778/hive-creating-hive-metastore-database-tables-command-fails-on-installation-pa