Elasticsearch: Update Settings API: Allow to dynamically change the number of replicas

Created on 8 Aug 2010 · 8Comments · Source: elastic/elasticsearch

A new API, update index / indices settings, allowing to change specific index level settings. The first setting supported is the index.number_of_replicas setting allowing to dynamically change the number of replicas an index has.

The REST endpoint is /_settings (to update all indices) or {index}/_settings to update one (or more) indices settings. The body of the request includes the updated settings, for example:

{
    "index" : {
        "number_of_replicas" : 4
    }
}

The above will change the number of replicas to 4 from the current number of replicas. Here is a curl example:

curl -XPUT 'localhost:9200/my_index/_settings' -d '
{
    "index" : {
        "number_of_replicas" : 4
    }
}
'

>feature v0.10.0

Source

kimchy

All 8 comments

implements.

kimchy on 8 Aug 2010

Nice! So when the number of replicas is increased, what happens? e.g.

Does ES start copying indices from master node to additional nodes?
Does replication start immediately?
Can one throttle the replication speed? (because sudden network and disk IO will affect query throughput)
Can one reduce the number or replicas?
In case of replica reduction, do indices get physically removed from disk?

Thanks.

otisg on 9 Aug 2010

👍1

Indices are not copied from the master node, they are recovered from a primary shard that is allocated "somewhere" in the cluster.
Yes.
There is built in throttling to not perform recovery from / to many shards at the same time.
Yes.
Yes.

kimchy on 9 Aug 2010

👍1

Btw, this isn't in the ES docs yet

clintongormley on 4 Sep 2010

Could you please tell how can we change index replica using java code???
I have done it through curl request but stuck while doing it through java code

ayush21792 on 28 Oct 2014

@ayush21792 you can find plenty of examples of using java API in elasticsearch integration tests. Saying that, the best place to ask questions like these is elasticsearch mailing list. We are trying to use github issues to track bugs and feature requests.

imotov on 31 Oct 2014

Thank you so much....
Could you please help me on another issue..When I index very large data of
12.7 million records,my elastic search server is terminating after 8.3
million records and showing the error of java heap memory overflow and the
error persists even after increasing memory.I am using 2gb java heap memory
and have also tried for 1gb and 3gb but result is same(halting at 8.3
million records).I have tried with both java api and es server both are
giving the same result.I am using windows 7.So could u please help me if
there is any setting which i am ignoring.

Thank You

On Fri, Oct 31, 2014 at 5:14 PM, Igor Motov [email protected]
wrote:

@ayush21792 https://github.com/ayush21792 you can find plenty of
examples of using java API in elasticsearch integration tests
https://github.com/elasticsearch/elasticsearch/blob/master/src/test/java/org/elasticsearch/indices/settings/UpdateNumberOfReplicasTests.java#L75.
Saying that, the best place to ask questions like these is elasticsearch
mailing list. We are trying to use github issues to track bugs and feature
requests.

—
Reply to this email directly or view it on GitHub
https://github.com/elasticsearch/elasticsearch/issues/302#issuecomment-61249220
.