Nicolas Favre-Felix created CASSANDRA-7531:
----------------------------------------------
Summary: Node replacement leads to null entries in system.peers
Key: CASSANDRA-7531
URL: https://issues.apache.org/jira/browse/CASSANDRA-7531
Project: Cassandra
Issue Type: Bug
Reporter: Nicolas Favre-Felix
On 2.0.9, I'm still reproducing a similar behavior as in CASSANDRA-7122.
One node was replaced and it's no longer in nodetool status but still has null
entries in system.peers:
{code}
peer | data_center | host_id |
preferred_ip | rack | release_version | rpc_address | schema_version
| tokens
---------------+-------------+--------------------------------------+--------------+---------+-----------------+---------------+--------------------------------------+--------------------------
10.123.99.36 | DC1 | null |
null | null | null | null |
null | null
10.123.45.121 | DC1 | e34b134c-c03e-464e-99a9-e3c022eed369 |
null | rack123 | 2.0.9-SNAPSHOT | 10.123.45.121 |
e6b32c6b-f3f3-3350-9bd1-a9b8620d42ea | {'4099276460824344804'}
10.123.77.88 | DC2 | fdd76fa0-0971-43d5-961b-452bd2821b31 |
null | rack456 | 2.0.9-SNAPSHOT | 10.123.77.88 |
e6b32c6b-f3f3-3350-9bd1-a9b8620d42ea | {'-3074457345618258603'}
[...]
{code}
This causes errors to be logged from the datastax Java driver (2.0.2):
{code}
[main] ERROR com.datastax.driver.core.ControlConnection - No rpc_address found
for host /10.123.99.36 in my.anonymized.host.name.com/10.123.45.66:9042's peers
system table. That should not happen but using address /10.123.99.36 instead
{code}
--
This message was sent by Atlassian JIRA
(v6.2#6252)
