如何将 600MB 大 json 文件批量插入到 elasticsearch？

Question

I am trying to insert 600MB Json file (which may be enlarge in the future) to elasticsearch. However, I get below error,

Error: "toString()" failed

I am using stream-json npm but no luck :( What is the best way to do this? I am thinking to chunk out, but if there's a better way, that'll be great

var makeBulk = function(csList, callback){
  const pipeline = fs.createReadStream('./CombinedServices_IBC.json').pipe(StreamValues.withParser());
  while()
  pipeline.on('data', data => {
    for(var index in data.value.features){
      bulk.push(
        { index: {_index: 'combinedservices1', _type: '_doc', _id: data.value.features[index].properties.OBJECTID } },
        {
          'geometry': data.value.features[index].geometry,
          'properties': data.value.features[index].properties
        }
      );
    }
    callback(bulk);

  });

}

Answer 1

There is a tool for such use case Elasticdump( https://github.com/taskrabbit/elasticsearch-dump )

Installation of elasticsearch-dump

npm install elasticdump -g
elasticdump

Import Json into ES

elasticdump \
  --input=./CombinedServices_IBC.json \
  --output=http://127.0.0.1:9200/my_index \
  --type=alias

Answer 2

Don't insert a bulk of 600MB, default bulk queue can keep up to 200 bulks inn JVM Heap Space - imagine if each is 600MB, what you will get is OOM and GC problems

Refer to https://www.elastic.co/guide/en/elasticsearch/guide/current/bulk.html#_how_big_is_too_big ; example logstash elasticsearch output plugin sends bulk of up to 20Mb