Avoid connection to nodes' internal IP addresses, preventing log warnings

[Lennart Liberg]

I'm running a cluster on google cloud, where the nodes communicate using their internal IP addresses.

When connecting to the cluster from external client using the the nodes' external IP addresses, these internal IPs gets returned as hosts, which makes the connection pool produce warnings since it can't connect to them.

In [2]: MyCluster.connect()

In [3]: WARNING:cassandra.pool:Error attempting to reconnect to 10.10.10.5, scheduling retry in 2.0 seconds: timed out

WARNING:cassandra.pool:Error attempting to reconnect to 10.10.10.9, scheduling retry in 2.0 seconds: timed out

WARNING:cassandra.pool:Error attempting to reconnect to 10.10.10.6, scheduling retry in 2.0 seconds: timed out

WARNING:cassandra.pool:Error attempting to reconnect to 10.10.10.9, scheduling retry in 4.0 seconds: timed out

WARNING:cassandra.pool:Error attempting to reconnect to 10.10.10.5, scheduling retry in 4.0 seconds: timed out

WARNING:cassandra.pool:Error attempting to reconnect to 10.10.10.6, scheduling retry in 4.0 seconds: timed out

In [26]: cqlengine.connection.session.hosts

Out[26]: 

[<Host: 104.10.10.2 euwestb>,

 <Host: 10.10.10.5 euwestb>,

 <Host: 146.10.10.9>,

 <Host: 10.10.10.8 euwestb>,

 <Host: 104.10.10.8>,

 <Host: 10.10.10.6 euwestb>,

 <Host: 104.10.10.1>]

What is a recommended way of avoiding these warnings while still maintaining the reconnect functionality for the accessible hosts?

(except of course setting a sufficiently high loglevel for the pool logger)?

Thanks,

Lennart Liberg

[Lennart Liberg]

Alright, the WhiteListRoundRobinPolicy seems to be the answer to my question.

Thanks!

To unsubscribe from this group and stop receiving emails from it, send an email to python-driver-u...@lists.datastax.com.