I have run the code successfully after I deleted the old ES version in
another machine.
What's the reason to this error even the ES in another machine is not
initialized?
Thanks for your time so much!!!
On Tue, Mar 5, 2013 at 6:05 PM, Jingang Wang bitwjg@gmail.com wrote:
I ever used ES 0.12 and now 0.90.
In my current machine, I run ES 0.90, I have not replace the old ES
version, but the ES instance in that machine is not initialized now.
would it be affected?
This error is mainly caused by mixing different ES versions, are you sure
you use only one ES version in the running nodes, on the local network?
Jörg
Am 05.03.13 10:53, schrieb Jingang Wang:
Thanks, but there occurs another exception. I think there should be some
mistakes in my settings of ES?
Exception in thread "main" org.elasticsearch.transport.**
TransportSerializationExceptio**n: Failed to deserialize exception
response from stream
at org.elasticsearch.transport.netty.MessageChannelHandler.
handlerResponseError(**MessageChannelHandler.java:**171)
--
Wang Jingang(王金刚)
Ph.D Candidate at
Lab of High Volume Language Information Processing & Cloud Computing
School of Computer Science
Beijing Institute of Technology
Beijing 100081
P.R China
--
Wang Jingang(王金刚)
Ph.D Candidate at
Lab of High Volume Language Information Processing & Cloud Computing
School of Computer Science
Beijing Institute of Technology
Beijing 100081
P.R China
I think there was another node up and running. You can check this by
looking at the hosts on the network. At ports 9300..9400, if ports are
open for listening, you can be sure there is an ES node active.
The EOFException is somehow unexpected when versions are mixed. But,
since the internal ES data structures change in a new version, they grow
in length. So if an old ES version responds with a short answer, the new
ES node waits for a longer version. But on the network layer, the
connection closes, and the new ES node gets notified with an EOFException.
For the future, Elasticsearch send magical version headers in all the
data transports which allow to detect the version of ES nodes which is
currently being talked to. So, new ES versions can deny data transport
more gracefully with a nice hint, and no more low level network
expections will happen.
Jörg
Am 05.03.13 11:09, schrieb Jingang Wang:
I have run the code successfully after I deleted the old ES version in
another machine.
What's the reason to this error even the ES in another machine is not
initialized?
Thanks for your time so much!!!
Thanks for your explanation.
Your demo code shows how to create index and query meanwhile after refresh.
What should I do if I want I create an Index with it active, and query it
when I need, because I need to add some documents into the index
ceaselessly.
Besides, I still do not find the error in my early code.
I think there was another node up and running. You can check this by
looking at the hosts on the network. At ports 9300..9400, if ports are open
for listening, you can be sure there is an ES node active.
The EOFException is somehow unexpected when versions are mixed. But, since
the internal ES data structures change in a new version, they grow in
length. So if an old ES version responds with a short answer, the new ES
node waits for a longer version. But on the network layer, the connection
closes, and the new ES node gets notified with an EOFException.
For the future, Elasticsearch send magical version headers in all the data
transports which allow to detect the version of ES nodes which is currently
being talked to. So, new ES versions can deny data transport more
gracefully with a nice hint, and no more low level network expections will
happen.
Jörg
Am 05.03.13 11:09, schrieb Jingang Wang:
I have run the code successfully after I deleted the old ES version in
another machine.
What's the reason to this error even the ES in another machine is not
initialized?
Thanks for your time so much!!!
--
Wang Jingang(王金刚)
Ph.D Candidate at
Lab of High Volume Language Information Processing & Cloud Computing
School of Computer Science
Beijing Institute of Technology
Beijing 100081
P.R China
Just put the query part elsewhere. You can consider using a river for
your indexing routine. A river is nothing but a plugin with a client
that fetches data from a source, and enters a loop for indexing data
contineously, e.g. from a file, a database, or a stream data source.
Jörg
Am 05.03.13 12:37, schrieb Jingang Wang:
What should I do if I want I create an Index with it active, and query
it when I need, because I need to add some documents into the index
ceaselessly.
I have resolved this problem, it was caused by different versions.
Because I ever used ES 0.20 before, so I referenced the jars in ES 0.20 in
my project. However, I forgot to re-reference these jars in my project
after I changed ES from 0.20 to 0.90.
The program could work after I resolved it.
I would express my sincerest thanks to both of your for your times and
great help.
Just put the query part elsewhere. You can consider using a river for your
indexing routine. A river is nothing but a plugin with a client that
fetches data from a source, and enters a loop for indexing data
contineously, e.g. from a file, a database, or a stream data source.
Jörg
Am 05.03.13 12:37, schrieb Jingang Wang:
What should I do if I want I create an Index with it active, and query it
when I need, because I need to add some documents into the index
ceaselessly.
--
Wang Jingang(王金刚)
Ph.D Candidate at
Lab of High Volume Language Information Processing & Cloud Computing
School of Computer Science
Beijing Institute of Technology
Beijing 100081
P.R China
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.
