Details
-
Bug
-
Status: Resolved
-
Low
-
Resolution: Fixed
-
linux
-
Low
Description
On a slow cluster (here used for testing purpose), cqlsh fails with a timeout error, whatever --connect-timeout option you can pass.
Here is a sample call:
cqlsh 192.168.XXX.YYY Connection error: ('Unable to connect to any servers', {'192.168.XXX.YYY': OperationTimedOut('errors=None, last_host=None',)})
cqlsh --connect-timeout=30 192.168.XXX.YYY Connection error: ('Unable to connect to any servers', {'192.168.XXX.YYY': OperationTimedOut('errors=None, last_host=None',)})
Debugging shows that the timeout is not properly propagated on the underlying ResponseWaiter.deliver() method in /usr/share/cassandra/lib/cassandra-driver-internal-only-3.0.0-6af642d.zip/cassandra-driver-3.0.0-6af642d/cassandra/connection.py
Workaround is to propagate, in cqlsh.py, the --connect-timeout option when initialize the cluster connection object (i.e. add kwarg "control_connection_timeout" in addition to the existing kwarg "connect_timeout")
Cluster( <other args>, control_connection_timeout=float(connect_timeout), connect_timeout=connect_timeout)