Delete by query APIedit

Deletes documents that match the specified query.

var deleteByQueryResponse = client.DeleteByQuery<Tweet>(d => d
    .Index("twitter")
    .Query(q => q
        .Match(m => m
            .Field(f => f.Message)
            .Query("some message")
        )
    )
);
POST /twitter/_delete_by_query
{
  "query": {
    "match": {
      "message": "some message"
    }
  }
}

Requestedit

POST /<index>/_delete_by_query

Descriptionedit

You can specify the query criteria in the request URI or the request body using the same syntax as the Search API.

When you submit a delete by query request, Elasticsearch gets a snapshot of the index when it begins processing the request and deletes matching documents using internal versioning. If a document changes between the time that the snapshot is taken and the delete operation is processed, it results in a version conflict and the delete operation fails.

Note

Documents with a version equal to 0 cannot be deleted using delete by query because internal versioning does not support 0 as a valid version number.

While processing a delete by query request, Elasticsearch performs multiple search requests sequentially to find all of the matching documents to delete. A bulk delete request is performed for each batch of matching documents. If a search or bulk request is rejected, the requests are retried up to 10 times, with exponential back off. If the maximum retry limit is reached, processing halts and all failed requests are returned in the response. Any delete requests that completed successfully still stick, they are not rolled back.

You can opt to count version conflicts instead of halting and returning by setting conflicts to proceeed.

Refreshing shardsedit

Specifying the refresh parameter refreshes all shards involved in the delete by query once the request completes. This is different than the delete API’s refresh parameter, which causes just the shard that received the delete request to be refreshed. Unlike the delete API, it does not support wait_for.

Running delete by query asynchronouslyedit

If the request contains wait_for_completion=false, Elasticsearch performs some preflight checks, launches the request, and returns a task you can use to cancel or get the status of the task. Elasticsearch creates a record of this task as a document at .tasks/task/${taskId}. When you are done with a task, you should delete the task document so Elasticsearch can reclaim the space.

Waiting for active shardsedit

wait_for_active_shards controls how many copies of a shard must be active before proceeding with the request. See Active shards for details. timeout controls how long each write request waits for unavailable shards to become available. Both work exactly the way they work in the Bulk API. Delete by query uses scrolled searches, so you can also specify the scroll parameter to control how long it keeps the search context alive, for example ?scroll=10m. The default is 5 minutes.

Throttling delete requestsedit

To control the rate at which delete by query issues batches of delete operations, you can set requests_per_second to any positive decimal number. This pads each batch with a wait time to throttle the rate. Set requests_per_second to -1 to disable throttling.

Throttling uses a wait time between batches so that the internal scroll requests can be given a timeout that takes the request padding into account. The padding time is the difference between the batch size divided by the requests_per_second and the time spent writing. By default the batch size is 1000, so if requests_per_second is set to 500:

target_time = 1000 / 500 per second = 2 seconds
wait_time = target_time - write_time = 2 seconds - .5 seconds = 1.5 seconds

Since the batch is issued as a single _bulk request, large batch sizes cause Elasticsearch to create many requests and wait before starting the next set. This is "bursty" instead of "smooth".

Slicingedit

Delete by query supports sliced scroll to parallelize the delete process. This can improve efficiency and provide a convenient way to break the request down into smaller parts.

Setting slices to auto chooses a reasonable number for most indices. If you’re slicing manually or otherwise tuning automatic slicing, keep in mind that:

  • Query performance is most efficient when the number of slices is equal to the number of shards in the index. If that number is large (for example, 500), choose a lower number as too many slices hurts performance. Setting slices higher than the number of shards generally does not improve efficiency and adds overhead.
  • Delete performance scales linearly across available resources with the number of slices.

Whether query or delete performance dominates the runtime depends on the documents being reindexed and cluster resources.

Path parametersedit

<index>
(Optional, string) A comma-separated list of index names to search. Use _all or omit to search all indices.

Query parametersedit

allow_no_indices

(Optional, boolean) If true, the request does not return an error if a wildcard expression or _all value retrieves only missing or closed indices.

This parameter also applies to index aliases that point to a missing or closed index.

analyzer
(Optional, string) Analyzer to use for the query string.
analyze_wildcard
(Optional, boolean) If true, wildcard and prefix queries are analyzed. Defaults to false.
conflicts
(Optional, string) What to do if delete by query hits version conflicts: abort or proceed. Defaults to abort.
default_operator
(Optional, string) The default operator for query string query: AND or OR. Defaults to OR.
df
(Optional, string) Field to use as default where no field prefix is given in the query string.
expand_wildcards

(Optional, string) Controls what kind of indices that wildcard expressions can expand to. Valid values are:

all
Expand to open and closed indices.
open
Expand only to open indices.
closed
Expand only to closed indices.
none
Wildcard expressions are not accepted.

Defaults to open.

from
(Optional, integer) Starting document offset. Defaults to 0.
ignore_unavailable
(Optional, boolean) If true, missing or closed indices are not included in the response. Defaults to false.
lenient
(Optional, boolean) If true, format-based query failures (such as providing text to a numeric field) will be ignored. Defaults to false.
max_docs
(Optional, integer) Maximum number of documents to process. Defaults to all documents.
preference
(Optional, string) Specifies the node or shard the operation should be performed on. Random by default.
q
(Optional, string) Query in the Lucene query string syntax.
request_cache
(Optional, boolean) If true, the request cache is used for this request. Defaults to the index-level setting.
refresh
(Optional, enum) If true, Elasticsearch refreshes the affected shards to make this operation visible to search, if wait_for then wait for a refresh to make this operation visible to search, if false do nothing with refreshes. Valid values: true, false, wait_for. Default: false.
requests_per_second
(Optional, integer) The throttle for this request in sub-requests per second. -1 means no throttle. Defaults to 0.
routing
(Optional, string) Target the specified primary shard.
scroll
(Optional, time units) Specifies how long a consistent view of the index should be maintained for scrolled search.
scroll_size
(Optional, integer) Size of the scroll request that powers the operation. Defaults to 100.
search_type

(Optional, string) The type of the search operation. Available options:

  • query_then_fetch
  • dfs_query_then_fetch
search_timeout
(Optional, time units) Explicit timeout for each search request. Defaults to no timeout.
slices
(Optional, integer) The number of slices this task should be divided into. Defaults to 1 meaning the task isn’t sliced into subtasks.
sort
(Optional, string) A comma-separated list of <field>:<direction> pairs.
_source
(Optional, string) True or false to return the _source field or not, or a list of fields to return.
_source_excludes
(Optional, string) A list of fields to exclude from the returned _source field.
_source_includes
(Optional, string) A list of fields to extract and return from the _source field.
stats
(Optional, string) Specific tag of the request for logging and statistical purposes.
terminate_after
(Optional, integer) The maximum number of documents to collect for each shard, upon reaching which the query execution will terminate early.
timeout
(Optional, time units) Specifies the period of time to wait for a response. If no response is received before the timeout expires, the request fails and returns an error. Defaults to 30s.
version
(Optional, boolean) If true, returns the document version as part of a hit.
wait_for_active_shards

(Optional, string) The number of shard copies that must be active before proceeding with the operation. Set to all or any positive integer up to the total number of shards in the index (number_of_replicas+1). Default: 1, the primary shard.

See Active shards.

Request bodyedit

query
(Optional, query object) Specifies the documents to delete using the Query DSL.

Response bodyedit

The JSON response looks like this:

{
  "took" : 147,
  "timed_out": false,
  "total": 119,
  "deleted": 119,
  "batches": 1,
  "version_conflicts": 0,
  "noops": 0,
  "retries": {
    "bulk": 0,
    "search": 0
  },
  "throttled_millis": 0,
  "requests_per_second": -1.0,
  "throttled_until_millis": 0,
  "failures" : [ ]
}
took
The number of milliseconds from start to end of the whole operation.
timed_out
This flag is set to true if any of the requests executed during the delete by query execution has timed out.
total
The number of documents that were successfully processed.
deleted
The number of documents that were successfully deleted.
batches
The number of scroll responses pulled back by the delete by query.
version_conflicts
The number of version conflicts that the delete by query hit.
noops
This field is always equal to zero for delete by query. It only exists so that delete by query, update by query, and reindex APIs return responses with the same structure.
retries
The number of retries attempted by delete by query. bulk is the number of bulk actions retried, and search is the number of search actions retried.
throttled_millis
Number of milliseconds the request slept to conform to requests_per_second.
requests_per_second
The number of requests per second effectively executed during the delete by query.
throttled_until_millis
This field should always be equal to zero in a _delete_by_query response. It only has meaning when using the Task API, where it indicates the next time (in milliseconds since epoch) a throttled request will be executed again in order to conform to requests_per_second.
failures
Array of failures if there were any unrecoverable errors during the process. If this is non-empty then the request aborted because of those failures. Delete by query is implemented using batches, and any failure causes the entire process to abort but all failures in the current batch are collected into the array. You can use the conflicts option to prevent reindex from aborting on version conflicts.

Examplesedit

Delete all tweets from the twitter index:

var deleteByQueryResponse = client.DeleteByQuery<Tweet>(d => d
    .Index("twitter")
    .Conflicts(Conflicts.Proceed)
    .Query(q => q.MatchAll())
);
POST twitter/_delete_by_query?conflicts=proceed
{
  "query": {
    "match_all": {}
  }
}

Delete documents from multiple indices:

var deleteByQueryResponse = client.DeleteByQuery<object>(d => d
    .Index("twitter,blog")
    .Query(q => q.MatchAll())
);
POST /twitter,blog/_delete_by_query
{
  "query": {
    "match_all": {}
  }
}

Limit the delete by query operation to shards that a particular routing value:

var deleteByQueryResponse = client.DeleteByQuery<Tweet>(d => d
    .Index("twitter")
    .Routing(1)
    .Query(q => q
        .Range(r => r
            .Field(f => f.Age)
            .GreaterThanOrEquals(10)
        )
    )
);
POST twitter/_delete_by_query?routing=1
{
  "query": {
    "range" : {
        "age" : {
           "gte" : 10
        }
    }
  }
}

By default _delete_by_query uses scroll batches of 1000. You can change the batch size with the scroll_size URL parameter:

var deleteByQueryResponse = client.DeleteByQuery<Tweet>(d => d
    .Index("twitter")
    .ScrollSize(5000)
    .Query(q => q
        .Term(r => r
            .Field(f => f.User)
            .Value("kimchy")
        )
    )
);
POST twitter/_delete_by_query?scroll_size=5000
{
  "query": {
    "term": {
      "user": "kimchy"
    }
  }
}
Slice manuallyedit

Slice a delete by query manually by providing a slice id and total number of slices:

var deleteByQueryResponse = client.DeleteByQuery<Tweet>(d => d
    .Index("twitter")
    .Slice(s => s
        .Id(0)
        .Max(2)
    )
    .Query(q => q
        .Range(r => r
            .Field(f => f.Likes)
            .LessThan(10)
        )
    )
);

var deleteByQueryResponse2 = client.DeleteByQuery<Tweet>(d => d
    .Index("twitter")
    .Slice(s => s
        .Id(1)
        .Max(2)
    )
    .Query(q => q
        .Range(r => r
            .Field(f => f.Likes)
            .LessThan(10)
        )
    )
);
POST twitter/_delete_by_query
{
  "slice": {
    "id": 0,
    "max": 2
  },
  "query": {
    "range": {
      "likes": {
        "lt": 10
      }
    }
  }
}
POST twitter/_delete_by_query
{
  "slice": {
    "id": 1,
    "max": 2
  },
  "query": {
    "range": {
      "likes": {
        "lt": 10
      }
    }
  }
}

Which you can verify works with:

var refreshResponse = client.Indices.Refresh();

var searchResponse = client.Search<Tweet>(s => s
    .Index("twitter")
    .Size(0)
    .FilterPath(new[] { "hits.total" }) 
    .Query(q => q
        .Range(r => r
            .Field(f => f.Likes)
            .LessThan(10)
        )
    )
);
GET _refresh
POST twitter/_search?size=0&filter_path=hits.total
{
  "query": {
    "range": {
      "likes": {
        "lt": 10
      }
    }
  }
}

Using filter path can result in a response that cannot be parsed by the client’s serializer. In these cases, using the low level client and parsing the JSON response may be preferred.

Which results in a sensible total like this one:

{
  "hits": {
    "total" : {
        "value": 0,
        "relation": "eq"
    }
  }
}
Use automatic slicingedit

You can also let delete-by-query automatically parallelize using sliced scroll to slice on _id. Use slices to specify the number of slices to use:

var deleteByQueryResponse = client.DeleteByQuery<Tweet>(d => d
    .Index("twitter")
    .Refresh()
    .Slices(5)
    .Query(q => q
        .Range(r => r
            .Field(f => f.Likes)
            .LessThan(10)
        )
    )
);
POST twitter/_delete_by_query?refresh&slices=5
{
  "query": {
    "range": {
      "likes": {
        "lt": 10
      }
    }
  }
}

Which you also can verify works with:

var searchResponse = client.Search<Tweet>(s => s
    .Index("twitter")
    .Size(0)
    .FilterPath(new[] { "hits.total" }) 
    .Query(q => q
        .Range(r => r
            .Field(f => f.Likes)
            .LessThan(10)
        )
    )
);
POST twitter/_search?size=0&filter_path=hits.total
{
  "query": {
    "range": {
      "likes": {
        "lt": 10
      }
    }
  }
}

Using filter path can result in a response that cannot be parsed by the client’s serializer. In these cases, using the low level client and parsing the JSON response may be preferred.

Which results in a sensible total like this one:

{
  "hits": {
    "total" : {
        "value": 0,
        "relation": "eq"
    }
  }
}

Setting slices to auto will let Elasticsearch choose the number of slices to use. This setting will use one slice per shard, up to a certain limit. If there are multiple source indices, it will choose the number of slices based on the index with the smallest number of shards.

Adding slices to _delete_by_query just automates the manual process used in the section above, creating sub-requests which means it has some quirks:

  • You can see these requests in the Tasks APIs. These sub-requests are "child" tasks of the task for the request with slices.
  • Fetching the status of the task for the request with slices only contains the status of completed slices.
  • These sub-requests are individually addressable for things like cancellation and rethrottling.
  • Rethrottling the request with slices will rethrottle the unfinished sub-request proportionally.
  • Canceling the request with slices will cancel each sub-request.
  • Due to the nature of slices each sub-request won’t get a perfectly even portion of the documents. All documents will be addressed, but some slices may be larger than others. Expect larger slices to have a more even distribution.
  • Parameters like requests_per_second and max_docs on a request with slices` are distributed proportionally to each sub-request. Combine that with the point above about distribution being uneven and you should conclude that using max_docs with slices might not result in exactly max_docs documents being deleted.
  • Each sub-request gets a slightly different snapshot of the source index though these are all taken at approximately the same time.
Change throttling for a requestedit

The value of requests_per_second can be changed on a running delete by query using the _rethrottle API. Rethrottling that speeds up the query takes effect immediately but rethrotting that slows down the query takes effect after completing the current batch to prevent scroll timeouts.

var rethrottleResponse = client.DeleteByQueryRethrottle("r1A2WoRbTwKZ516z6NEs5A:36619",
    r => r
    .RequestsPerSecond(-1)
);
POST _delete_by_query/r1A2WoRbTwKZ516z6NEs5A:36619/_rethrottle?requests_per_second=-1

Use the tasks API to get the task ID. Set requests_per_second to any positive decimal value or -1 to disable throttling.

Get the status of a delete by query operationedit

Use the tasks API to get the status of a delete by query operation:

var listTasksResponse = client.Tasks.List(t => t
    .Detailed()
    .Actions("*/delete/byquery")
);
GET _tasks?detailed=true&actions=*/delete/byquery

The response looks like:

{
  "nodes" : {
    "r1A2WoRbTwKZ516z6NEs5A" : {
      "name" : "r1A2WoR",
      "transport_address" : "127.0.0.1:9300",
      "host" : "127.0.0.1",
      "ip" : "127.0.0.1:9300",
      "attributes" : {
        "testattr" : "test",
        "portsfile" : "true"
      },
      "tasks" : {
        "r1A2WoRbTwKZ516z6NEs5A:36619" : {
          "node" : "r1A2WoRbTwKZ516z6NEs5A",
          "id" : 36619,
          "type" : "transport",
          "action" : "indices:data/write/delete/byquery",
          "status" : {    
            "total" : 6154,
            "updated" : 0,
            "created" : 0,
            "deleted" : 3500,
            "batches" : 36,
            "version_conflicts" : 0,
            "noops" : 0,
            "retries": 0,
            "throttled_millis": 0
          },
          "description" : ""
        }
      }
    }
  }
}

This object contains the actual status. It is just like the response JSON with the important addition of the total field. total is the total number of operations that the reindex expects to perform. You can estimate the progress by adding the updated, created, and deleted fields. The request will finish when their sum is equal to the total field.

With the task id you can look up the task directly:

var getTaskResponse = client.Tasks.GetTask("r1A2WoRbTwKZ516z6NEs5A:36619");
GET /_tasks/r1A2WoRbTwKZ516z6NEs5A:36619

The advantage of this API is that it integrates with wait_for_completion=false to transparently return the status of completed tasks. If the task is completed and wait_for_completion=false was set on it then it’ll come back with results or an error field. The cost of this feature is the document that wait_for_completion=false creates at .tasks/task/${taskId}. It is up to you to delete that document.

Cancel a delete by query operationedit

Any delete by query can be canceled using the task cancel API:

var cancelTaskResponse = client.Tasks.Cancel(t => t
    .TaskId("r1A2WoRbTwKZ516z6NEs5A:36619")
);
POST _tasks/r1A2WoRbTwKZ516z6NEs5A:36619/_cancel

The task ID can be found using the tasks API.

Cancellation should happen quickly but might take a few seconds. The task status API above will continue to list the delete by query task until this task checks that it has been cancelled and terminates itself.