[
https://issues.apache.org/jira/browse/CASSANDRA-6663?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13898877#comment-13898877
]
Sylvain Lebresne commented on CASSANDRA-6663:
---------------------------------------------
bq. https://github.com/xerial/snappy-java/issues/37
So let me sum this up: when executing the OPTIONS message, which asks for which
compression methods are available, Snappy segfault on raspberry PI and we we
can't catch that (since it segfault the JVM). So really, the "right" solution
is probably to help on that snappy-java tracking the source of the problem down
so it gets fixed there.
But in meantime, if there was a simple and reliable way to detect that "we are
on raspberry PI", we could probably just mark Snappy as unavailable on
raspberry without actually trying to initialize it. But for that, someone that
does have a raspberry PI will have to step up with a patch because I don't
personally have one.
Lastly, the python driver could probably work around that by not sending the
OPTIONS message if the user hasn't asked for compression in the first place
(not saying it's ideal, but currently there isn't much reason to send OPTIONS
unless for knowing which compressions are available, so it could be a
relatively simple work-around until the snappy-java library ticket is fixed).
> Connecting to a Raspberry PI Cassandra Cluster crashes the node being
> connected to
> ----------------------------------------------------------------------------------
>
> Key: CASSANDRA-6663
> URL: https://issues.apache.org/jira/browse/CASSANDRA-6663
> Project: Cassandra
> Issue Type: Bug
> Components: Drivers (now out of tree)
> Environment: 4x node Raspberry PI cluster
> Macbook running Idle 2.7
> Reporter: ian mccrae
> Attachments: Python Client Log.txt, hs_err_pid6327.log
>
>
> I have a working 4x node Raspberry Pi cluster and
> # DevCenter happily connects to this (...which has an option to turn Snappy
> compression off)
> # ...however the Python Driver fails to connect and crashes the node being
> connected to with the errors in the error-log below.
> There appears to be a problem with Snappy compression (not supported on the
> Raspberry Pi). So I also tried "compression = None" with the same result.
> How might I fix this?
> *Python Code*
> {noformat}
> >>> from cassandra.cluster import Cluster
> >>> cluster = Cluster(['192.168.200.151'], compression = None)
> >>> session = cluster.connect()
> {noformat}
> *Error Log*
> {noformat}
> Traceback (most recent call last):
> File "<pyshell#58>", line 1, in <module>
> session = cluster.connect()
> File
> "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/cassandra/cluster.py",
> line 471, in connect
> self.control_connection.connect()
> File
> "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/cassandra/cluster.py",
> line 1351, in connect
> self._set_new_connection(self._reconnect_internal())
> File
> "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/cassandra/cluster.py",
> line 1386, in _reconnect_internal
> raise NoHostAvailable("Unable to connect to any servers", errors)
> NoHostAvailable: ('Unable to connect to any servers', {'192.168.200.151':
> ConnectionShutdown('Connection to 192.168.200.151 is closed',))
> {noformat}
> *A Dump of the "cluster" class attributes*
> {noformat}
> >>> pprint(vars(cluster))
> {'_core_connections_per_host': {0: 2, 1: 1},
> '_is_setup': True,
> '_is_shutdown': True,
> '_listener_lock': <thread.lock object at 0x10616d230>,
> '_listeners': set([]),
> '_lock': <_RLock owner=None count=0>,
> '_max_connections_per_host': {0: 8, 1: 2},
> '_max_requests_per_connection': {0: 100, 1: 100},
> '_min_requests_per_connection': {0: 5, 1: 5},
> '_prepared_statements': <WeakValueDictionary at 4396942904>,
> 'compression': None,
> 'contact_points': ['192.168.200.151'],
> 'control_connection': <cassandra.cluster.ControlConnection object at
> 0x106168cd0>,
> 'control_connection_timeout': 2.0,
> 'cql_version': None,
> 'executor': <concurrent.futures.thread.ThreadPoolExecutor object at
> 0x106148410>,
> 'load_balancing_policy': <cassandra.policies.RoundRobinPolicy object at
> 0x104adae50>,
> 'max_schema_agreement_wait': 10,
> 'metadata': <cassandra.metadata.Metadata object at 0x1061481d0>,
> 'metrics_enabled': False,
> 'port': 9042,
> 'scheduler': <cassandra.cluster._Scheduler object at 0x106148550>,
> 'sessions': <_weakrefset.WeakSet object at 0x106148750>,
> 'sockopts': None,
> 'ssl_options': None}
> >>>
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
