I'm running version 0.90.3 on a 6 node ES cluster. My first Java API based
program to index log entries from a file works fine. So I'm trying to morph
it to run via TCP socket connection to accept "live" feeds via Redis. The
program is running on the same workstation - which is one of the cluster
nodes. I am using this code, which works fine in the file-based app:
But when I attempt to create the client, I get this error:
13/10/03 14:10:57 INFO elasticsearch.node: [t5] version[0.90.3], pid[6505],
build[5c38d60/2013-08-06T13:18:31Z]
13/10/03 14:10:57 INFO elasticsearch.node: [t5] initializing ...
13/10/03 14:10:57 INFO elasticsearch.plugins: [t5] loaded [], sites [head]
13/10/03 14:10:59 ERROR state.meta: [t5] failed to read local state,
exiting...
org.elasticsearch.ElasticSearchIllegalStateException: node is not
configured to store local location
All searching for a cause/solution seem to refer to something called
"graylog" (which I am not using), and none
of them have an answer.
you use NodeClient (there is also TransportClient which in my eyes is
simpler to use)
a NodeClient uses cluster discovery behind the scenes
because a NodeClient is a full member of the cluster, it must be able to
read and save cluster state via local gateway
in your example, there is no setting for configuration of the path for
the local gateway, so the default settings apply
your node client wants to synchronize with the cluster by sending its own
state, but can't read a previously saved state, which is fatal, and then it
bails out
Thank you for all the info Jorg. I'm not sure I understand it all, but will
switch to TransportClient and see how that goes. I'm still confused that
the file-based program works, but the environment must somehow be different.
Terry
On Thursday, October 3, 2013 2:38:39 PM UTC-4, Jörg Prante wrote:
Just a quick helper
you use NodeClient (there is also TransportClient which in my eyes is
simpler to use)
a NodeClient uses cluster discovery behind the scenes
because a NodeClient is a full member of the cluster, it must be able to
read and save cluster state via local gateway
in your example, there is no setting for configuration of the path for
the local gateway, so the default settings apply
your node client wants to synchronize with the cluster by sending its
own state, but can't read a previously saved state, which is fatal, and
then it bails out
TransportClient worked like a charm - thanks Jorg.
On Thursday, October 3, 2013 2:48:29 PM UTC-4, Terry Healy wrote:
Thank you for all the info Jorg. I'm not sure I understand it all, but
will switch to TransportClient and see how that goes. I'm still confused
that the file-based program works, but the environment must somehow be
different.
Terry
On Thursday, October 3, 2013 2:38:39 PM UTC-4, Jörg Prante wrote:
Just a quick helper
you use NodeClient (there is also TransportClient which in my eyes is
simpler to use)
a NodeClient uses cluster discovery behind the scenes
because a NodeClient is a full member of the cluster, it must be able
to read and save cluster state via local gateway
in your example, there is no setting for configuration of the path for
the local gateway, so the default settings apply
your node client wants to synchronize with the cluster by sending its
own state, but can't read a previously saved state, which is fatal, and
then it bails out
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.