[
https://issues.apache.org/jira/browse/CASSANDRA-3775?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jonathan Ellis resolved CASSANDRA-3775.
---------------------------------------
Resolution: Not A Problem
This is expected behavior. Cassandra only sends one data request for each
query, so if that replica is dead-but-not-detected-dead then the request will
time out.
> rpc_timeout error when reading from a cluster that just had a node die. Only
> happens if gossip hasn't noticed the dead node yet.
> --------------------------------------------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-3775
> URL: https://issues.apache.org/jira/browse/CASSANDRA-3775
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Affects Versions: 1.0.6
> Environment: ubuntu. Used ccm to create the cluster.
> Reporter: Tyler Patterson
>
> Create a cluster of 3 nodes with RF=3 and CL=QUORUM. Insert some data then
> kill a node (not the coordinator) and immediately try to read the data. The
> read request will fail within about 2 seconds. cassandra.yaml has
> rpc_timeout=10000. A failing test has been written in cassandra-dtest, branch
> "read_when_node_is_down". The test can be run like this: nosetests
> --nocapture read_when_node_down_test.py Here is the error from the test:
> {code}
> ======================================================================
> ERROR: read_when_node_down_test.TestReadWhenNodeDown.read_when_node_down_test
> ----------------------------------------------------------------------
> Traceback (most recent call last):
> File "/usr/lib/pymodules/python2.7/nose/case.py", line 187, in runTest
> self.test(*self.arg)
> File "/home/tahooie/cassandra-dtest/read_when_node_down_test.py", line 40,
> in read_when_node_down_test
> query_c1c2(cursor, 100, CL)
> File "/home/tahooie/cassandra-dtest/tools.py", line 28, in query_c1c2
> cursor.execute('SELECT c1, c2 FROM cf USING CONSISTENCY %s WHERE key=k%d'
> % (consistency, key))
> File "/usr/local/lib/python2.7/dist-packages/cql/cursor.py", line 96, in
> execute
> raise cql.OperationalError("Request did not complete within rpc_timeout.")
> OperationalError: ('Request did not complete within rpc_timeout.', 'reading
> failed in 2.0130 seconds.')
> {code}
> I did notice that if I sleep 20 seconds after killing the node and before
> reading, that the read succeeds. This is probably because gossip has had a
> chance to notice that the node is down.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
