Configuration Management
This chapter lists the static configuration items that can be modified in the console and the configuration items that can be dynamically updated through API.
The console supports modifying common static configuration parameters to the elasticsearch.yml configuration file. These configurations do not support dynamic updates and need to restart the cluster after modification.
| Configuration item | Default value | Configuration description |
|---|---|---|
| cluster.name | instance_id | Cluster name |
| discovery.zen.ping.unicast.hosts | [] | Unicast address, default master node IP or the first three node IPs, example [127.0.0.1,127.0.0.2,127.0.0.3] |
| gateway.recover_after_nodes | N - 2 | Recovery can be performed when the expected number of nodes is reached |
| gateway.expected_nodes | N | Expected number of cluster nodes. When the expected number of nodes join the cluster, shard recovery will begin |
| gateway.recover_after_time | 5m | If the expected number of nodes is not reached, no attempt to recovery, and the recovery process will wait the configured time |
| http.cors.enabled | true | Allow cross-domain access |
| http.cors.allow-origin | * | Domains that allow cross-domain access by default support all domains |
| rest.action.multi.allow_explicit_index | true | Allow the index parameter in the body to override the index parameter in the URL |
| node.attr.tag | Node tag, not set by default | |
| gateway.expected_master_nodes | 0 | Expected number of master nodes in the cluster. When the expected number of masters join the cluster, shard recovery will begin |
| gateway.expected_data_nodes | 0 | Expected number of data nodes in the cluster. When the expected number of data nodes join the cluster, shard recovery will begin |
| gateway.recover_after_master_nodes | 0 | Recovery can be performed when the expected number of master nodes is reached |
| gateway.recover_after_data_nodes | 0 | Recovery can be performed when the expected number of data nodes is reached |
| http.max_content_length | 100mb | Maximum content of the HTTP request, default 100mb. If set to greater than Integer.MAX_VALUE, it will be reset to 100mb |
| http.max_initial_line_length | 4kb | Maximum length of the HTTP URL, default 4kb |
| http.max_header_size | 8kb | Maximum value of the allowed headers, default 8kb |
| http.compression | true | Supports compression as much as possible (using Accept-Encoding) |
| http.compression_level | 3 | Defines the compression level of the HTTP response. Valid values range between 1 (minimum compression) and 9 (maximum compression) |
| http.cors.max-age | 1728000 | Browser sends “preflight” requests to determine CORS settings. Max-age defines how long the result should be cached. Default is 20 days |
| http.cors.allow-methods | OPTIONS, HEAD, GET, POST, PUT, DELETE | Allowed methods |
| http.cors.allow-headers | X-Requested-With, Content-Type, Content-Length | Allowed headers |
| http.cors.allow-credentials | false | Whether the Access-Control-Allow-Credentials header should be returned. Note: This header will only be returned when set to true |
| http.detailed_errors.enabled | true | Enable or disable the output of detailed error messages and stack traces in response output. Note: If set to false, and the error_trace request parameter is specified, an error will be returned; when error_trace is not specified, a simple message will be returned |
| http.pipelining | true | Enable or disable HTTP pipelining |
| http.pipelining.max_events | 10000 | Maximum number of events queued in memory before an HTTP connection closes |
| indices.fielddata.cache.size | unlimited | Maximum value of field data cache, for example, 30% of node heap space, or absolute value, like 12GB |
| indices.queries.cache.size | 10% | Controls the memory size of the filter cache, default 10%. Accepts percentage values (such as 5%) or precise values (such as 512mb) |
| index.queries.cache.enabled | true | Controls whether to enable the query cache, the setting can be on the basis of each configurable index setting |
| indices.memory.index_buffer_size | 10% | Accepts percentage or byte size value. Default 10%, meaning that 10% of the total heap allocated to the node will be used as the shared index buffer across all shards |
| indices.memory.min_index_buffer_size | 48mb | If specifying index_buffer_size as a percentage, you can use this setting to specify the absolute minimum |
| indices.memory.max_index_buffer_size | unlimited | If you specify index_buffer_size as a percentage, you can use this setting to specify the absolute maximum |
| indices.requests.cache.size | 1% | Cache is managed at the node level, default maximum is 1% of heap |
| node.ingest | true | Ingest qualifications |
| search.remote.connect | true | Cross-cluster search |
| index.number_of_shards | 5 | Number of primary shards an index should have, this setting can be dynamically set when creating the index |
| index.shard.check_on_startup | false | Whether to check for damage before opening the shard; when damage is detected, it will prevent the shard from being opened |
| index.routing_partition_size | 1 | Number of shards a custom routing value can go to, default 1, can only be set when creating the index. This value must be less than index.number_of_shards unless the value of index.number_of_shards is also 1 |
- Note: N is the number of nodes
Most of the configuration items not listed above can be dynamically modified through cluster-update-settings. For a detailed description of the dynamic configuration items, please refer to Modules
Below are common dynamically updatable configuration parameters.
| Configuration item | Default value | Configuration description |
|---|---|---|
| cluster.routing.allocation.enable | all | Enable or disable allocation of specific kinds of shards |
| cluster.routing.allocation.node_concurrent_incoming_recoveries | 2 | How many concurrent incoming shard recoveries are allowed to occur on a node |
| cluster.routing.allocation.node_concurrent_outgoing_recoveries | 2 | How many concurrent outgoing shard recoveries are allowed to occur on a node |
| cluster.routing.allocation.node_concurrent_recoveries | 2 | Shortcut setting for node_concurrent_incoming_recoveries and node_concurrent_outgoing_recoveries |
| cluster.routing.allocation.node_initial_primaries_recoveries | 4 | After the node starts, recovery of unassigned master nodes will use data on the local disk |
| cluster.routing.allocation.same_shard.host | false | Allows to check for allocating multiple instances of the same shard on a single host |
| cluster.routing.rebalance.enable | all | Enable or disable rebalancing of specific kinds of shards |
| cluster.routing.allocation.allow_rebalance | indices_all_active | Specifies when shard rebalancing is allowed |
| cluster.routing.allocation.cluster_concurrent_rebalance | 2 | Allows control of how many concurrent shard rebalances are allowed for the cluster-wide |
| cluster.routing.allocation.balance.shard | 0.45f | Defines the weight factor for the total number of shards allocated on the node (floating point number) |
| cluster.routing.allocation.balance.index | 0.55f | Defines the weight factor for the number of shards of each index allocated on a specific node (floating point number) |
| cluster.routing.allocation.balance.threshold | 1.0f | The minimum optimization value (non-negative floating point number) for the action should be performed |
| cluster.routing.allocation.disk.threshold_enabled | true | Set to false to disable the disk allocation decider |
| cluster.routing.allocation.disk.watermark.low | 85％ | Controls the low watermark of disk usage, default 85％. Beyond that, ES will not allocate new shards to the node |
| cluster.routing.allocation.disk.watermark.high | 90% | Controls the high watermark, default 90％. Beyond that, ES will try to relocate shards to another node |
| cluster.routing.allocation.disk.watermark.flood_stage | 95% | Controls flood stage, default 95％, beyond which ES forces a read-only index block on each index of one or multiple shards allocated on a node where at least one disk is exceeded. Once there is enough disk space available for index operations to continue, the index block must be manually released |
| cluster.info.update.interval | 30s | ES should check the disk usage of each node in the cluster |
| cluster.routing.allocation.disk.include_relocations | true | ES will consider the shards currently being relocated to the target node when calculating the disk usage of the node |
| cluster.routing.allocation.awareness.attributes.* | Shard allocation awareness settings allow you to tell ES about your hardware configuration | |
| cluster.routing.allocation.include.{attribute} | Allocate shards to nodes with at least one comma-separated value for {attribute}, attribute could be _name, _ip, _host | |
| cluster.routing.allocation.require.{attribute} | Only assign shards to nodes that have all comma-separated values for {attribute}, attribute might be _name, _ip, _host | |
| cluster.routing.allocation.exclude.{attribute} | Don’t assign shards to nodes without the comma-separated values for {attribute}, attribute might be _name, _ip, _host | |
| cluster.blocks.read_only | false | Makes the entire cluster read-only (index does not accept write operations), data is not allowed to be modified (create or delete the index) |
| cluster.blocks.read_only_allow_delete | false | The same as cluster.blocks.read_only, but allows index deletion to free up resources |
| cluster.indices.tombstones.size | 500 | The cluster status maintains index tombstones to explicitly represent deleted indexes |
| logger.org.elasticsearch.indices.recovery | INFO | Log recording level |
| discovery.zen.ping.unicast.hosts.resolve_timeout | 5s | Wait time for DNS lookup before each round of ping |
| discovery.zen.ping_timeout | 3s | Ping timeout time |
| discovery.zen.join_timeout | 60s | Default timeout is 20 times the ping timeout |
| discovery.zen.minimum_master_nodes | N/2 + 1 | Minimum number of master node qualifications, N is the number of master nodes |
| discovery.zen.no_master_block | write | A setting that controls which operations should be rejected when there is no active master node, default is write |
| indices.breaker.total.limit | 70% | Starting limit of the overall parent breaker, default is 70% of JVM heap |
| indices.breaker.fielddata.limit | 60% | Limit of the fielddata breaker, default is 60% of JVM heap |
| indices.breaker.fielddata.overhead | 1.03 | A constant, all fielddata estimates are multiplied to decide the final estimate |
| indices.breaker.request.limit | 60% | Limit of the request breaker, default is 60% of JVM heap |
| indices.breaker.request.overhead | 1 | A constant, all request estimates are multiplied to decide the final estimate |
| network.breaker.inflight_requests.limit | 100% | Limit of the inflight_requests breaker, default is 100% of JVM heap |
| network.breaker.inflight_requests.overhead | 1 | A constant, all inflight requests estimates are multiplied to decide the final |
| script.max_compilations_rate | 75/5m | Limit of the number of unique dynamic scripts allowed to be compiled within a certain interval |
| index.requests.cache.enable | true | Enable or disable index cache |
| indices.recovery.max_bytes_per_sec | 40mb | Maximum bandwidth for data transmission between nodes |
| indices.store.throttle.max_bytes_per_sec | 20mb | Maximum bandwidth for writing to disk |
| index.merge.scheduler.max_thread_count | Maximum number of threads for index merge, default is max(1, min(4, availableProcessors / 2)) | |
| index.number_of_replicas | 1 | Number of replicas each primary shard has |
| index.auto_expand_replicas | default is false (i.e, disabled) | Automatically expands the number of replicas according to the number of available nodes. Set to a range delimited by a hyphen for the lower and upper limit (e.g., 0-5) |
| index.refresh_interval | 1s | How often to perform a refresh operation, which makes the recent changes to the index visible for search. Can be set to -1 to disable refresh |
| index.max_result_window | 10000 | The maximum value of from + size for searching the index |
| index.max_inner_result_window | 100 | The maximum value of from + size for internal hits defined and top aggregated to this index |
| index.max_rescore_window | 10000 | Maximum window_size for rescore requests used for searching this index. Default is index.max_result_window |
| index.max_docvalue_fields_search | 100 | Maximum number of docvalue_fields_allowed in a query |
| index.max_script_fields | 32 | Maximum number of script_fields allowed in a query |
| index.max_ngram_diff | 1 | Maximum allowed difference between min_gram and max_gram for NGramTokenizer and NGramTokenFilter |
| index.max_shingle_diff | 3 | Maximum allowed difference between max_shingle_size and min_shingle_size |
| index.blocks.read_only | false | Set to true to make the index and index data read-only, otherwise allow write and data changes |
| index.blocks.read_only_allow_delete | false | The same as index.blocks.read_only, but allows index deletion to free up resources |
| index.blocks.read | false | Set to true to disallow read operations on the index |
| index.blocks.write | false | Set to true to disallow write operations on the index |
| index.blocks.metadata | false | Set to true to disable reading and writing of index data |
| index.max_refresh_listeners | The maximum number of refresh listeners available on each shard of the index. These listeners are used to implement refresh = wait_for |