Shrinking doesn't play well with index.mapping.single_type

[jpountz]

Here is a recreation for 5.6 (only 5.x is affected, not 6.x where this setting doesn't make sense):

DELETE test

PUT test 
{
  "settings": {
    "number_of_shards": 2,
    "mapping.single_type": true
  },
  "mappings": {
    "doc": {
      "properties": {
        "foo": {
          "type": "keyword"
        }
      }
    }
  }
}

PUT test/doc/1
{
  "foo": "bar"
}

PUT /test/_settings
{
  "settings": {
    "index.blocks.write": true 
  }
}

POST test/_shrink/test2

GET test2/_search


{
  "error": {
    "root_cause": [
      {
        "type": "illegal_state_exception",
        "reason": "Call postProcess before getting the uid"
      }
    ],
    "type": "search_phase_execution_exception",
    "reason": "all shards failed",
    "phase": "query",
    "grouped": true,
    "failed_shards": [
      {
        "shard": 0,
        "index": "test2",
        "node": "hKTfMOBARL-EapXTxdKcUA",
        "reason": {
          "type": "illegal_state_exception",
          "reason": "Call postProcess before getting the uid"
        }
      }
    ]
  },
  "status": 500
}

This is due to the fact that the single type setting is not preserved by the shrink:

GET test2/_settings 

{
  "test2" : {
    "settings" : {
      "index" : {
        "routing" : {
          "allocation" : {
            "initial_recovery" : {
              "_id" : "hKTfMOBARL-EapXTxdKcUA"
            }
          }
        },
        "allocation" : {
          "max_retries" : "1"
        },
        "number_of_shards" : "1",
        "routing_partition_size" : "1",
        "shrink" : {
          "source" : {
            "name" : "test",
            "uuid" : "RPfBwJwFTteDPXgcmZlzbw"
          }
        },
        "provided_name" : "test2",
        "creation_date" : "1530698402187",
        "number_of_replicas" : "1",
        "uuid" : "ql1DbgPsTs6oKuIIl-BsCQ",
        "version" : {
          "created" : "5061199",
          "upgraded" : "5061199"
        }
      }
    }
  }
}

[elasticmachine]

Pinging @elastic/es-core-infra

[jpountz]

@gmoskovicz just let me know that we actually already fixed it in 6.3 via #29202 cc @jimczi

[jimczi]

Yep this is fixed in 6.3 but I don't remember why I did not backport in 5.6 😕 . I'll open a new pr for 5.6, sorry for the confusion.

[jimczi]

Closed by #31811, thanks @jpountz and @gmoskovicz !