Im trying to connect externally to a containerized distributed apache drill setup on DC/OS using the JDBC driver as shown here
https://drill.apache.org/docs/using-the-jdbc-driver/#example-of-connecting-to-drill-programmatically
However, when running this outside of the cluster to connect from my machine, the zookeeper connection completes, then it tries to resolve a Drillbit address in the form of
2d7f7217340c:31010
where 2d7f7217340c is the internal drillbit id as shown in the drill web console
which of course throws
java.nio.channels.UnresolvedAddressException
until I make an entry in my local /etc/hosts file
to resolve 2d7f7217340c to the public ip of the drill node.
I've searched around quite a bit on the topic and not found anything. Is there a way to configure Zookeeper or Drill to return an IP rather than the drillbit ID so that the hosts file doesn't need to be manually updated?
I noticed that the address string 2d7f7217340c is defined by the container ID generated during deployment. Even with modifying the hostname of the container via
lsns
nsenter --target <container_pid> --uts hostname <public_ip>
and restarting drill in the container, zookeeper still returns the 2d7f7217340c address.
A working solution is to modify the recent addition of the DRILL_HOSTNAME variable in apache-drill/conf/drill-env.sh to grab the public hostname via the AWS metadata service
export DRILL_HOSTNAME=`curl -s http://169.254.169.254/latest/meta-data/public-hostname`
the public hostname is now advertised by Drill, and can be resolved both externally and internally