I'd like to create a new index on some time interval (day or week)
where all indexes have the same format, in order to grow the total
number of shards over time and allow for the total number of servers
in the cluster to grow over time. This happens nicely and
automatically if I just add a date stamp to my index name and continue
indexing records as I get them. Is there a way to define a mapping
that will be inherited by indexes that are created automatically by
Elastic Search?
There isn't an option to do that currently, the ability for an index created
to automatically get a set of mappings applied to it. Open an issue for it,
should be simple to implement 
-shay.banon
On Wed, Mar 24, 2010 at 8:52 PM, Matt matthew.maul@gmail.com wrote:
I'd like to create a new index on some time interval (day or week)
where all indexes have the same format, in order to grow the total
number of shards over time and allow for the total number of servers
in the cluster to grow over time. This happens nicely and
automatically if I just add a date stamp to my index name and continue
indexing records as I get them. Is there a way to define a mapping
that will be inherited by indexes that are created automatically by
Elastic Search?
Great. My thought was that something like:
curl -XPUT http://localhost:9200/_all/object_name/_mapping -d pay_load
would apply the mapping to current indexes, as well as future indexes.
Does that make sense? Should I phrase the issue as an enhancement to
multi-index mapping, or a totally new feature?
On Mar 24, 3:00 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
There isn't an option to do that currently, the ability for an index created
to automatically get a set of mappings applied to it. Open an issue for it,
should be simple to implement-shay.banon
On Wed, Mar 24, 2010 at 8:52 PM, Matt matthew.m...@gmail.com wrote:
I'd like to create a new index on some time interval (day or week)
where all indexes have the same format, in order to grow the total
number of shards over time and allow for the total number of servers
in the cluster to grow over time. This happens nicely and
automatically if I just add a date stamp to my index name and continue
indexing records as I get them. Is there a way to define a mapping
that will be inherited by indexes that are created automatically by
Elastic Search?
Actually, thats a great idea, I did not think about exposing it as an API,
but allowing to specify a list of mappings to load when started. Yours is
much better. Just create an issue as a new feature.
I need to think about the syntax though, since you can already do _all,
which applies it to the current indices (and not future ones, as its not
implemented yet ).
-shay.banon
On Wed, Mar 24, 2010 at 9:14 PM, Matt matthew.maul@gmail.com wrote:
Great. My thought was that something like:
curl -XPUT http://localhost:9200/_all/object_name/_mapping -d pay_load
would apply the mapping to current indexes, as well as future indexes.
Does that make sense? Should I phrase the issue as an enhancement to
multi-index mapping, or a totally new feature?On Mar 24, 3:00 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
There isn't an option to do that currently, the ability for an index
created
to automatically get a set of mappings applied to it. Open an issue for
it,
should be simple to implement-shay.banon
On Wed, Mar 24, 2010 at 8:52 PM, Matt matthew.m...@gmail.com wrote:
I'd like to create a new index on some time interval (day or week)
where all indexes have the same format, in order to grow the total
number of shards over time and allow for the total number of servers
in the cluster to grow over time. This happens nicely and
automatically if I just add a date stamp to my index name and continue
indexing records as I get them. Is there a way to define a mapping
that will be inherited by indexes that are created automatically by
Elastic Search?
Your way would be nice too. If I could store my mappings in a config
file and have them apply to all future indexes, I wouldn't have to
clutter my application code at all; the mappings would just be picked
up by the cluster at start-up time. As an API, it's sort of an extra
manual task when you boot a cluster to define the mappings, but it
would allow for adding a new type of object to a growing set of
indices. Maybe I'll create a ticket for each and let you decide what's
a better idea.
On Mar 24, 4:03 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
Actually, thats a great idea, I did not think about exposing it as an API,
but allowing to specify a list of mappings to load when started. Yours is
much better. Just create an issue as a new feature.I need to think about the syntax though, since you can already do _all,
which applies it to the current indices (and not future ones, as its not
implemented yet-shay.banon
On Wed, Mar 24, 2010 at 9:14 PM, Matt matthew.m...@gmail.com wrote:
Great. My thought was that something like:
curl -XPUThttp://localhost:9200/_all/object_name/_mapping-d pay_load
would apply the mapping to current indexes, as well as future indexes.
Does that make sense? Should I phrase the issue as an enhancement to
multi-index mapping, or a totally new feature?On Mar 24, 3:00 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
There isn't an option to do that currently, the ability for an index
created
to automatically get a set of mappings applied to it. Open an issue for
it,
should be simple to implement-shay.banon
On Wed, Mar 24, 2010 at 8:52 PM, Matt matthew.m...@gmail.com wrote:
I'd like to create a new index on some time interval (day or week)
where all indexes have the same format, in order to grow the total
number of shards over time and allow for the total number of servers
in the cluster to grow over time. This happens nicely and
automatically if I just add a date stamp to my index name and continue
indexing records as I get them. Is there a way to define a mapping
that will be inherited by indexes that are created automatically by
Elastic Search?
cool, I will try and work on it soon.
-shay.banon
On Wed, Mar 24, 2010 at 10:20 PM, Matt matthew.maul@gmail.com wrote:
Your way would be nice too. If I could store my mappings in a config
file and have them apply to all future indexes, I wouldn't have to
clutter my application code at all; the mappings would just be picked
up by the cluster at start-up time. As an API, it's sort of an extra
manual task when you boot a cluster to define the mappings, but it
would allow for adding a new type of object to a growing set of
indices. Maybe I'll create a ticket for each and let you decide what's
a better idea.On Mar 24, 4:03 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
Actually, thats a great idea, I did not think about exposing it as an
API,
but allowing to specify a list of mappings to load when started. Yours is
much better. Just create an issue as a new feature.I need to think about the syntax though, since you can already do _all,
which applies it to the current indices (and not future ones, as its not
implemented yet-shay.banon
On Wed, Mar 24, 2010 at 9:14 PM, Matt matthew.m...@gmail.com wrote:
Great. My thought was that something like:
curl -XPUThttp://localhost:9200/_all/object_name/_mapping-d pay_load
would apply the mapping to current indexes, as well as future indexes.
Does that make sense? Should I phrase the issue as an enhancement to
multi-index mapping, or a totally new feature?On Mar 24, 3:00 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
There isn't an option to do that currently, the ability for an index
created
to automatically get a set of mappings applied to it. Open an issue
for
it,
should be simple to implement-shay.banon
On Wed, Mar 24, 2010 at 8:52 PM, Matt matthew.m...@gmail.com
wrote:I'd like to create a new index on some time interval (day or week)
where all indexes have the same format, in order to grow the total
number of shards over time and allow for the total number of
servers
in the cluster to grow over time. This happens nicely and
automatically if I just add a date stamp to my index name and
continue
indexing records as I get them. Is there a way to define a mapping
that will be inherited by indexes that are created automatically by
Elastic Search?
Awesome. Thanks for being so quick to respond.
On Mar 24, 5:28 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
cool, I will try and work on it soon.
-shay.banon
On Wed, Mar 24, 2010 at 10:20 PM, Matt matthew.m...@gmail.com wrote:
Your way would be nice too. If I could store my mappings in a config
file and have them apply to all future indexes, I wouldn't have to
clutter my application code at all; the mappings would just be picked
up by the cluster at start-up time. As an API, it's sort of an extra
manual task when you boot a cluster to define the mappings, but it
would allow for adding a new type of object to a growing set of
indices. Maybe I'll create a ticket for each and let you decide what's
a better idea.On Mar 24, 4:03 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
Actually, thats a great idea, I did not think about exposing it as an
API,
but allowing to specify a list of mappings to load when started. Yours is
much better. Just create an issue as a new feature.I need to think about the syntax though, since you can already do _all,
which applies it to the current indices (and not future ones, as its not
implemented yet-shay.banon
On Wed, Mar 24, 2010 at 9:14 PM, Matt matthew.m...@gmail.com wrote:
Great. My thought was that something like:
curl -XPUThttp://localhost:9200/_all/object_name/_mapping-dpay_load
would apply the mapping to current indexes, as well as future indexes.
Does that make sense? Should I phrase the issue as an enhancement to
multi-index mapping, or a totally new feature?On Mar 24, 3:00 pm, Shay Banon shay.ba...@elasticsearch.com wrote:
There isn't an option to do that currently, the ability for an index
created
to automatically get a set of mappings applied to it. Open an issue
for
it,
should be simple to implement-shay.banon
On Wed, Mar 24, 2010 at 8:52 PM, Matt matthew.m...@gmail.com
wrote:I'd like to create a new index on some time interval (day or week)
where all indexes have the same format, in order to grow the total
number of shards over time and allow for the total number of
servers
in the cluster to grow over time. This happens nicely and
automatically if I just add a date stamp to my index name and
continue
indexing records as I get them. Is there a way to define a mapping
that will be inherited by indexes that are created automatically by
Elastic Search?