illegal_argument_exception: [routing] is missing for join field [joinField]

• ElasticDump version: 3.3.19
• Elasticsearch version: 6.3.0
• Node.js version (note that nodejs v8+ is required for this tool): v9.11.2
• Full command you are having issue with (with inputs and outputs)

Exporting to file (works):

#!/usr/bin/env bash
elasticdump \
  --input=http://10.0.0.1:9200/index \
  --output=/data/index-mapping.json \
  --type=mapping

elasticdump \
  --input=http://10.0.0.1:9200/index \
  --output=/data/index-analyzer.json \
  --type=analyzer

elasticdump \
  --input=http://10.0.0.1:9200/index \
  --output=/data/index-data.json \
  --type=data

importing from file on another server (doesn’t work)

#!/usr/bin/env bash
elasticdump \
  --input=index-analyzer.json \
  --output=http://127.0.0.1:9200/index \
  --type=analyzer

elasticdump \
  --input=index-mapping.json \
  --output=http://127.0.0.1:9200/index \
  --type=mapping

elasticdump \
  --input=index-data.json \
  --output=http://127.0.0.1:9200/index \
  --type=data

It fails with:

{ _index: 'index',
  _type: 'doc',
  _id: 'article-112437',
  status: 400,
  error:
   { type: 'mapper_parsing_exception',
     reason: 'failed to parse',
     caused_by:
      { type: 'illegal_argument_exception',
        reason: '[routing] is missing for join field [joinField]' } } }

Currently working on a simple data set to reproduce.

[tholu]

@ferronrsmith Can you verify the analysis of @todvora and reopen the issue? Thanks!

[tholu]

I can confirm the issue for the following versions:

~# elasticdump --version
4.1.1

~# curl localhost:9200
{
  "name" : "S74eeM1",
  "cluster_name" : "docker-cluster",
  "cluster_uuid" : "WlNA_Dn-Roy8dKNwW0GXlg",
  "version" : {
    "number" : "6.4.2",
    "build_flavor" : "default",
    "build_type" : "tar",
    "build_hash" : "04711c2",
    "build_date" : "2018-09-26T13:34:09.098244Z",
    "build_snapshot" : false,
    "lucene_version" : "7.4.0",
    "minimum_wire_compatibility_version" : "5.6.0",
    "minimum_index_compatibility_version" : "5.0.0"
  },
  "tagline" : "You Know, for Search"
}

The dataset that causes this is around 30GB, so still working on providing a simple dataset to reproduce.