Hazel Bobrins created CASSANDRA-12753:
-----------------------------------------
Summary: Create MV can corrupt C*, blocking all further table
actions and startup
Key: CASSANDRA-12753
URL: https://issues.apache.org/jira/browse/CASSANDRA-12753
Project: Cassandra
Issue Type: Bug
Components: Core
Environment: RHEL6.5
Cas 3.0.9
Reporter: Hazel Bobrins
Priority: Critical
Attachments: Cass_start.txt, MV_Create.txt, table_drop.txt
Creating a MV with a protected field name e.g. 'set' results in an error. Post
this failed MV create all further actions in the keyspace fail and node startup
fails until the keyspace is dropped.
Tested on a fresh 3.0.9 install single node cluster.
How to reproduce
cassandra@cqlsh:test1> CREATE KEYSPACE test1 WITH replication = {'class':
'SimpleStrategy', 'replication_factor': 1 } AND durable_writes = 'true';
cassandra@cqlsh:test1> use test1 ;
cassandra@cqlsh:test1> CREATE TABLE main_table ( field1 text, field2 text,
"set" text, PRIMARY KEY ( field1, field2 ) );
cassandra@cqlsh:test1> CREATE MATERIALIZED VIEW mv1 AS SELECT field2, field1,
"set" FROM main_table WHERE field1 IS NOT NULL AND field2 IS NOT NULL PRIMARY
KEY ( field2, field1 ) ;
ServerError: java.lang.RuntimeException:
java.util.concurrent.ExecutionException:
org.apache.cassandra.exceptions.SyntaxException: line 1:23 no viable
alternative at input 'set' (SELECT field1, field2, [set]...)
## Attached stack traces - 'MV_Create' thrown at this point
cassandra@cqlsh:test1> drop TABLE main_table ;
ServerError: java.lang.RuntimeException:
java.util.concurrent.ExecutionException:
org.apache.cassandra.exceptions.SyntaxException: line 1:23 no viable
alternative at input 'set' (SELECT field1, field2, [set]...)
## Attached stacke traces - 'Table_drop' thrown at this point
Finally restart Cassandra. Attach stack 'Cass_start' thrown at this point and
C* does not start.
Dropping the keyspace does work, however, this must obviously be done before
stopping the node.
We have also tested this on a 4 node cluster, post the MV create all nodes
report the same issue.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
