Breaking changes in 7.4edit

This section discusses the changes that you need to be aware of when migrating your application to Elasticsearch 7.4.

See also What’s new in 7.17 and Release notes.

Plugins changesedit

TokenizerFactory changesedit

TokenizerFactory now has a name() method that must be implemented. Most plugin-provided TokenizerFactory implementations will extend AbstractTokenizerFactory, which now takes a name parameter in its constructor.

Search Changesedit

Forbid empty doc values in vector functionsedit

If a document doesn’t have a value for a vector field (dense_vector or sparse_vector) on which a vector function is executed, an error will be thrown.

Use float instead of double for query vectorsedit

Previously, vector functions like cosineSimilarity represented the query vector as an list of doubles. Now vector functions use floats, which matches how the stored document vectors are represented.

Snapshot and Restore changesedit

The S3 repository plugin uses the DNS style access pattern by defaultedit

Starting in version 7.4 the repository-s3 plugin does not use the now-deprecated path-style access pattern by default. In versions 7.0, 7.1, 7.2 and 7.3 the repository-s3 plugin always used the path-style access pattern. This is a breaking change for deployments that only support path-style access but which are recognized as supporting DNS-style access by the AWS SDK. If your deployment only supports path-style access and is affected by this change then you must configure the S3 client setting path_style_access to true. This breaking change was made necessary by AWS’s announcement that the path-style access pattern is deprecated and will be unsupported on buckets created after September 30th 2020.

HTTP changesedit

Changes to Encoding Plus Signs in URLsedit

Starting in version 7.4, a + in a URL will be encoded as %2B by all REST API functionality. Prior versions handled a + as a single space. If your application requires handling + as a single space you can return to the old behaviour by setting the system property to true. Note that this behaviour is deprecated and setting this system property to true will cease to be supported in version 8.

Cluster changesedit

Rerouting after starting a shard runs at lower priorityedit

After starting each shard the elected master node must perform a reroute to search for other shards that could be allocated. In particular, when creating an index it is this task that allocates the replicas once the primaries have started. In versions prior to 7.4 this task runs at priority URGENT, but starting in version 7.4 its priority is reduced to NORMAL. In a well-configured cluster this reduces the amount of work the master must do, but means that a cluster with a master that is overloaded with other tasks at HIGH or URGENT priority may take longer to allocate all replicas.

Additionally, before 7.4 the GET _cluster_health?wait_for_no_initializing_shards and GET _cluster/health?wait_for_no_relocating_shards APIs would return only once all pending reroutes have completed too, but starting in version 7.4 if you want to wait for the rerouting process to completely finish you should add the wait_for_events=languid query parameter when calling these APIs.

Allocation changesedit

Auto-release of read-only-allow-delete blockedit

If a node exceeds the flood-stage disk watermark then Elasticsearch adds the index.blocks.read_only_allow_delete block to all of its indices to prevent further writes, as a last-resort attempt to prevent the node completely exhausting its disk space. In earlier versions this block would remain in place until manually removed, causing confusion for users who currently have ample disk space and who are not aware that they nearly ran out at some point in the past. From 7.4 onwards the block is automatically removed when the node drops below the high watermark again, with the expectation that the high watermark is some distance below the flood-stage watermark and therefore the disk space problem is truly resolved. Since this block may be automatically removed, you can no longer rely on adding this block manually to prevent writes to an index. You should use the index.blocks.read_only block instead. This behaviour can be disabled by setting the system property es.disk.auto_release_flood_stage_block to false.

Settings changesedit

Authentication realm name uniqueness is enforcededit

Authentication realm name uniqueness is now enforced. If you configure more than one realm of any type with the same name, the node fails to start.

pidfile setting is being replaced by node.pidfileedit

To ensure that all settings are in a proper namespace, the pidfile setting is deprecated, and will be removed in version 8.0.0. Instead, use node.pidfile.

processors setting is being replaced by node.processorsedit

To ensure that all settings are in a proper namespace, the processors setting is deprecated, and will be removed in version 8.0.0. Instead, use node.processors.

Transform changesedit

Stats response format changesedit

The response format of the Get transform statistics is very different to previous versions:

  • task_state and indexer_state are combined into a single state field that replaces the old state object.
  • Within the checkpointing object, current is renamed to last and in_progress to next.
  • The checkpoint number is now nested under last and next.
  • checkpoint_progress is now reported in an object nested in the next checkpoint object. (If there is no next checkpoint then no checkpoint is in progress and by definition the last checkpoint is 100% complete.)

For an example of the new format see Examples.

Data frame analytics changesedit

Changes to progress reportingedit

The single integer progress_percent field at the top level of the data frame analytics job stats is replaced by a progress field that is an array of objects. Each object contains the phase name and progress_percent of one phase of the analytics. For example:

  "id" : "my_job",
  "state" : "analyzing",
  "progress" : [
      "phase" : "reindexing",
      "progress_percent" : 100
      "phase" : "loading_data",
      "progress_percent" : 100
      "phase" : "analyzing",
      "progress_percent" : 47
      "phase" : "writing_results",
      "progress_percent" : 0