I just found out, that it is possible to mappings as files in the
filesystem. I like that feature very much, and the operations
department even more
Is there any possibility to store the index configuration (which is
done on index creation) as well on the filesystem? I did not find any
documentation, or more likely searched wrong.
Is this a) possible and b) a good idea except that I like it
I just found out, that it is possible to mappings as files in the
filesystem. I like that feature very much, and the operations
department even more
Is there any possibility to store the index configuration (which is
done on index creation) as well on the filesystem? I did not find any
documentation, or more likely searched wrong.
Is this a) possible and b) a good idea except that I like it
sorry for not being clear. I know this works. I meant whether the same
is possible for index configuration (analyzers etc) and not only for
the mappings.
I just found out, that it is possible to mappings as files in the
filesystem. I like that feature very much, and the operations
department even more
Is there any possibility to store the index configuration (which is
done on index creation) as well on the filesystem? I did not find any
documentation, or more likely searched wrong.
Is this a) possible and b) a good idea except that I like it
To answer to point b)
Not sure that is a best practice. I think (not sure about that) that if you put settings in files, you will have to copy it on each node.
Sending settings through curl will copy settings on each node.
I just found out, that it is possible to mappings as files in the
filesystem. I like that feature very much, and the operations
department even more
Is there any possibility to store the index configuration (which is
done on index creation) as well on the filesystem? I did not find any
documentation, or more likely searched wrong.
Is this a) possible and b) a good idea except that I like it
This is true. we just load ours up from a json file in our rails app when
we need indexes with different shard/replica settings. And then we have a
default setting that is in the elasticsearch.yml file that is used when an
index is created w/o specified settings.
Regards,
Dan
On Tue, Jul 12, 2011 at 6:49 AM, David Pilato david@pilato.fr wrote:
To answer to point b)
Not sure that is a best practice. I think (not sure about that) that if you
put settings in files, you will have to copy it on each node.
Sending settings through curl will copy settings on each node.
I just found out, that it is possible to mappings as files in the
filesystem. I like that feature very much, and the operations
department even more
Is there any possibility to store the index configuration (which is
done on index creation) as well on the filesystem? I did not find any
documentation, or more likely searched wrong.
Is this a) possible and b) a good idea except that I like it
On Tue, 2011-07-12 at 14:49 +0200, David Pilato wrote:
To answer to point b)
Not sure that is a best practice. I think (not sure about that) that if you put settings in files, you will have to copy it on each node.
Sending settings through curl will copy settings on each node.
Correct me if I'm wrong...
Absolutely correct.
The danger with using config files is that you need to be sure that all
your nodes have the same config.
To my mind, much easier to do it in your application via the API.
if you are starting with a single node it is still somewhat handy to
have all configuration loaded on startup instead of calling the HTTP
API before being able to pull data in.
In case of a clustered setup you are absolutely right of course.
Regarding index level settings, you can put the settings in the elasticsearch main configuration, but then it will apply to all indices created. You can put mappings per index (or across all indices), in the file system, but not index settings. Its a missing feature. Though I prefer people will configure elasticsearch using an API to make it simpler when they move to a distributed environment, you can open a feature to allow to specify index level settings in the file system, shouldn't be hard to implement.
On Tuesday, July 12, 2011 at 4:14 PM, Alexander Reelsen wrote:
Hi there,
if you are starting with a single node it is still somewhat handy to
have all configuration loaded on startup instead of calling the HTTP
API before being able to pull data in.
In case of a clustered setup you are absolutely right of course.
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.