Configuration settings

Configures the general policy for when checkpoints should occur. Possible values are:

One of [PERIODIC, CONTINUOUS, VOLUME, VOLUMETRIC].

Configures the time interval between checkpoints. The database does not checkpoint more often than the specified interval (unless checkpointing is triggered by a different event) but might checkpoint less often if performing a checkpoint takes longer time than the configured interval. A checkpoint is a point in the transaction logs from which recovery starts. Longer checkpoint intervals typically mean that recovery takes longer to complete in case of a crash. On the other hand, a longer checkpoint interval can also reduce the I/O load that the database places on the system, as each checkpoint implies a flushing and forcing of all the store files.

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s).

Configures the transaction interval between checkpoints. The database does not checkpoint more often than the specified interval (unless checkpointing is triggered by a different event) but might checkpoint less often if performing a checkpoint takes longer time than the configured interval. A checkpoint is a point in the transaction logs from which recovery starts. Longer checkpoint intervals typically mean that recovery takes longer to complete in case of a crash. On the other hand, a longer checkpoint interval can also reduce the I/O load that the database places on the system, as each checkpoint implies a flushing and forcing of all the store files. The default is 100000 for a checkpoint every 100000 transactions.

An integer that is minimum 1.

Configures the volume of transaction logs between checkpoints. The database does not checkpoint more often than the specified interval (unless checkpointing is triggered by a different event) but might checkpoint less often if performing a checkpoint takes longer time than the configured interval. A checkpoint is a point in the transaction logs, which recovery would start from. Longer checkpoint intervals typically mean that recovery takes longer to complete in case of a crash. On the other hand, a longer checkpoint interval can also reduce the I/O load that the database places on the system, as each checkpoint implies a flushing and forcing of all the store files.

A byte size (valid multipliers are B, KiB, KB, K, kB, kb, k, MiB, MB, M, mB, mb, m, GiB, GB, G, gB, gb, g, TiB, TB, PiB, PB, EiB, EB) that is minimum 1.00KiB.

Advertised address for this connector.

A socket address in the format of hostname:port, hostname, or :port that is an accessible address. If missing, it is acquired from server.default_advertised_address.

The maximum time to wait before sending a NOOP on connections waiting for responses from active ongoing queries.The minimum value is 1 millisecond.

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s) that is minimum 1ms.

The type of messages to enable keep-alive messages for ALL, STREAMING, or OFF.

One of [ALL, STREAMING, OFF].

The total number of probes to be missed before a connection is considered stale. The minimum value is 1.

An integer that is minimum 1.

The interval between every scheduled keep-alive check on all connections with active queries. Zero duration turns off keep-alive service.

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s) that is minimum 0s.

Enable the Bolt connector.

A boolean.

Address the connector should bind to.

A socket address in the format of hostname:port, hostname, or :port. If missing, it is acquired from server.default_listen_address.

Enable server OCSP stapling for bolt and http connectors.

A boolean.

The maximum time an idle thread in the thread pool bound to this connector waits for new tasks.

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s).

The maximum number of threads allowed in the thread pool bound to this connector.

An integer.

The number of threads, including idle, to keep in the thread pool bound to this connector.

An integer.

The encryption level to be used to secure communications with this connector.

One of [REQUIRED, OPTIONAL, DISABLED].

Advertised address for this connector.

A socket address in the format of hostname:port, hostname, or :port that is an accessible address. If missing, it is acquired from server.default_advertised_address.

Enable the HTTP connector.

A boolean.

Address the connector should bind to.

A socket address in the format of hostname:port, hostname, or :port. If missing, it is acquired from server.default_listen_address.

Defines the set of modules loaded into the Neo4j web server. The enterprise management endpoints are only available in the Еnterprise edition.

A comma-separated set where each element is one of [TRANSACTIONAL_ENDPOINTS, UNMANAGED_EXTENSIONS, BROWSER, ENTERPRISE_MANAGEMENT_ENDPOINTS, QUERY_API_ENDPOINTS].

Advertised address for this connector.

A socket address in the format of hostname:port, hostname, or :port that is an accessible address. If missing, it is acquired from server.default_advertised_address.

Enable the HTTPS connector.

A boolean.

Address the connector should bind to.

A socket address in the format of hostname:port, hostname, or :port. If missing, it is acquired from server.default_listen_address.

Default hostname or IP address the server uses to advertise itself.

A socket address in the format of hostname:port, hostname, or :port that has no specified port and is an accessible address.

Default network interface to listen for incoming connections. To listen for connections on all interfaces, use "0.0.0.0".

A socket address in the format of hostname:port, hostname, or :port that has no specified port.

Address routing connector should bind to.

A socket address in the format of hostname:port, hostname, or :port. If missing, it is acquired from server.default_listen_address.

Routing strategy for neo4j:// protocol connections. Default is CLIENT, using client-side routing, with server-side routing as a fallback (if enabled). When set to SERVER, client-side routing is short-circuited, and requests rely on server-side routing (which must be enabled for proper operation, i.e. dbms.routing.enabled=true). Can be overridden by dbms.routing.client_side.enforce_for_domains.

One of [SERVER, CLIENT].

Socket connection timeout. A timeout of zero is treated as an infinite timeout and will be bound by the timeout configured on the operating system level.

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s).

Pooled connections older than this threshold will be closed and removed from the pool. Setting this option to a low value will cause a high connection churn and might result in a performance hit. It is recommended to set maximum lifetime to a slightly smaller value than the one configured in network equipment (load balancer, proxy, firewall, etc. can also limit maximum connection lifetime). Zero and negative values result in lifetime not being checked.

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s).

Maximum amount of time spent attempting to acquire a connection from the connection pool. This timeout only kicks in when all existing connections are being used, and no new connections can be created because the maximum connection pool size has been reached. An error is raised when no connection can be acquired within the configured time. Negative values are allowed, which results in an unlimited acquisition timeout. A value of 0 is allowed, resulting in no timeout and immediate failure when the connection is unavailable.

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s).

Pooled connections that have been idle in the pool for longer than this timeout will be tested to ensure they are still alive before being used again. If the value of this option is too low, acquiring a connection will require an additional network call, which will cause a performance hit. If the value of this option is too high, live connections might no longer be used, leading to errors. Hence, this parameter balances the likelihood of experiencing connection problems and performance. Usually, this parameter should not need tuning. Value 0 means connections will always be tested for validity. No connection liveliness check is done by default.

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s).

Maximum total number of connections to be managed by a connection pool. The limit is enforced for a combination of a host and user. Negative values are allowed and result in unlimited pool. Value of 0 is not allowed. Defaults to -1 (unlimited).

An integer.

Sets the level for the driver’s internal logging.

One of [DEBUG, INFO, WARN, ERROR, NONE].

Enable server-side routing in clusters using an additional bolt connector. When configured, this allows requests to be forwarded from one cluster member to another, if the requests cannot be satisfied by the first member (e.g. write requests received by a non-leader).

A boolean.

How long callers should cache the response of the routing procedure dbms.routing.getRoutingTable().

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s) that is minimum 1s.

This setting is associated with performance optimization. Set this to true in situations where it is preferable to have any queries using the 'shortestPath' function terminate as soon as possible with no answer, rather than potentially running for a long time attempting to find an answer (even if there is no path to be found). For most queries, the 'shortestPath' algorithm will return the correct answer very quickly. However there are some cases where it is possible that the fast bidirectional breadth-first search algorithm will find no results even if they exist. This can happen when the predicates in the WHERE clause applied to 'shortestPath' cannot be applied to each step of the traversal, and can only be applied to the entire path. When the query planner detects these special cases, it will plan to perform an exhaustive depth-first search if the fast algorithm finds no paths. However, the exhaustive search may be orders of magnitude slower than the fast algorithm. If it is critical that queries terminate as soon as possible, it is recommended that this option be set to true, which means that Neo4j will never consider using the exhaustive search for shortestPath queries. However, please note that if no paths are found, an error will be thrown at run time, which will need to be handled by the application.

A boolean.

This setting is associated with performance optimization. The shortest path algorithm does not work when the start and end nodes are the same. With this setting set to false no path will be returned when that happens. The default value of true will instead throw an exception. This can happen if you perform a shortestPath search after a cartesian product that might have the same start and end nodes for some of the rows passed to shortestPath. If it is preferable to not experience this exception, and acceptable for results to be missing for those rows, then set this to false. If you cannot accept missing results, and really want the shortestPath between two common nodes, then re-write the query using a standard Cypher variable length pattern expression followed by ordering by path length and limiting to one result.

A boolean.

Set this to specify the behavior when Cypher planner or runtime hints cannot be fulfilled. If true, then non-conformance will result in an error, otherwise only a warning is generated.

A boolean.

Allow label inference during cardinality estimation. If the planner can logically deduce that a node has a label not explicitly expressed in the query, the planner will use this information during cardinality estimation.
This setting controls to what extent the planner should do that:

One of [MOST_SELECTIVE_LABEL, OFF].

Set this to change the behavior for Cypher create relationship when the start or end node is missing. By default this fails the query and stops execution, but by setting this flag the create operation is simply not performed and execution continues.

A boolean.

The minimum time between possible Cypher query replanning events. After this time, the graph statistics will be evaluated, and if they have changed by more than the value set by dbms.cypher.statistics_divergence_threshold, the query will be replanned. If the statistics have not changed sufficiently, the same interval will need to pass before the statistics will be evaluated again. Each time they are evaluated, the divergence threshold will be reduced slightly until it reaches 10% after 7h, so that even moderately changing databases will see query replanning after a sufficiently long time interval.

A duration (Valid units are: ns, μs, ms, s, m, h and d; default unit is s).

Set this to specify the default planner for the default language version.

One of [DEFAULT, COST].

The threshold for statistics above which a plan is considered stale.

If any of the underlying statistics used to create the plan have changed more than this value, the plan will be considered stale and will be replanned. Change is calculated as abs(a-b)/max(a,b).

This means that a value of 0.75 requires the database to quadruple in size before query replanning. A value of 0 means that the query will be replanned as soon as there is any change in statistics and the replan interval has elapsed.

This interval is defined by dbms.cypher.min_replan_interval and defaults to 10s. After this interval, the divergence threshold will slowly start to decline, reaching 10% after about 7h. This will ensure that long running databases will still get query replanning on even modest changes, while not replanning frequently unless the changes are very large.

A double that is in the range 0.0 to 1.0.

Allows the enabling or disabling of the file watcher service. This is an auxiliary service but should be left enabled in almost all cases.

A boolean.

Relationship count threshold for considering a node to be dense.

An integer that is minimum 1.

Specify if Neo4j should try to preallocate store files as they grow.

A boolean.

Database timezone for temporal functions. All Time and DateTime values that are created without an explicit timezone will use this configured default timezone.

A string describing a timezone, either described by offset (e.g. +02:00) or by name (e.g. Europe/Stockholm).

Name of the default database (aliases are not supported).
NOTE: This setting is not the same as dbms.default_database, which was used to set the default database in Neo4j 4.x and earlier versions.

The initial.dbms.default_database setting is meant to set the default database before the creation of the DBMS. Once it is created, the setting is not valid anymore.

To set the default database, use the dbms.setDefaultDatabase() procedure instead.

A valid database name containing only alphabetic characters, numbers, dots, and dashes with a length between 3 and 63 characters, starting with an alphabetic character but not with the name system.

Database timezone. Among other things, this setting influences the monitoring procedures.

One of [UTC, SYSTEM].

The size of the internal buffer in bytes used by LOAD CSV. If the csv file contains huge fields this value may have to be increased.

A long that is minimum 1.

Selects whether to conform to the standard https://tools.ietf.org/html/rfc4180 for interpreting escaped quotation characters in CSV files loaded using LOAD CSV. Setting this to false will use the standard, interpreting repeated quotes '""' as a single in-lined quote, while true will use the legacy convention originally supported in Neo4j 3.0 and 3.1, allowing a backslash to include quotes in-lined in fields.

A boolean.

The name of the analyzer that the full-text indexes should use by default.

A string.

Whether or not full-text indexes should be eventually consistent by default or not.

A boolean.

The eventually consistent mode of the full-text indexes works by queueing up index updates to be applied later in a background thread. This newBuilder sets an upper bound on how many index updates are allowed to be in this queue at any one point in time. When it is reached, the commit process will slow down and wait for the index update applier thread to make some more room in the queue.

An integer that is in the range 1 to 50000000.

Enable or disable background index sampling.

A boolean.

Index sampling chunk size limit.

An integer that is in the range 1048576 to 2147483647.

Percentage of index updates of total index size required before sampling of a given index is triggered.

An integer that is minimum 0.

Enable HTTP request logging.

A boolean.

Path to the logging configuration for debug, query, http and security logs.

A path. If relative, it is resolved from server.directories.neo4j_home.

Enable the debug log.

A boolean.

Enable GC Logging.

A boolean.

GC Logging Options.

A string.

Number of GC logs to keep.

An integer.

Size of each GC log that is kept.

A byte size (valid multipliers are B, KiB, KB, K, kB, kb, k, MiB, MB, M, mB, mb, m, GiB, GB, G, gB, gb, g, TiB, TB, PiB, PB, EiB, EB).

Path to the logging configuration of user logs.

A path. If relative, it is resolved from server.directories.neo4j_home.

Page cache can be configured to perform usage sampling of loaded pages that can be used to construct active load profile. According to that profile pages can be reloaded on the restart, replication, etc. This setting allows disabling that behavior. This feature is available in Neo4j Enterprise Edition.

A boolean.

Page cache warmup can be configured to prefetch files, preferably when cache size is bigger than store size. Files to be prefetched can be filtered by 'dbms.memory.pagecache.warmup.preload.allowlist'. Enabling this disables warmup by profile.

A boolean.

Page cache warmup prefetch file allowlist regex. By default matches all files.

A string.

Enable off heap and on heap memory tracking. Should not be set to false for clusters.

A boolean.

Initial heap size. By default it is calculated based on available system resources.

A byte size (valid multipliers are B, KiB, KB, K, kB, kb, k, MiB, MB, M, mB, mb, m, GiB, GB, G, gB, gb, g, TiB, TB, PiB, PB, EiB, EB).

Maximum heap size. By default it is calculated based on available system resources.

A byte size (valid multipliers are B, KiB, KB, K, kB, kb, k, MiB, MB, M, mB, mb, m, GiB, GB, G, gB, gb, g, TiB, TB, PiB, PB, EiB, EB).

Defines the size of the off-heap memory blocks cache. The cache will contain this number of blocks for each block size that is power of two. Thus, maximum amount of memory used by blocks cache can be calculated as 2 * server.memory.off_heap.max_cacheable_block_size * server.memory.off_heap.block_cache_size

An integer that is minimum 16.

Defines the maximum size of an off-heap memory block that can be cached to speed up allocations. The value must be a power of 2.

A byte size (valid multipliers are B, KiB, KB, K, kB, kb, k, MiB, MB, M, mB, mb, m, GiB, GB, G, gB, gb, g, TiB, TB, PiB, PB, EiB, EB) that is minimum 4.00KiB and is power of 2.

Use direct I/O for page cache. This setting is supported only on Linux and only for a subset of record formats that use platform-aligned page size.

A boolean.

The maximum number of worker threads to use for pre-fetching data when doing sequential scans. Set to '0' to disable pre-fetching for scans.

An integer that is in the range 0 to 255.

The amount of memory to use for mapping the store files. If Neo4j is running on a dedicated server, then it is generally recommended to leave about 2-4 gigabytes for the operating system, give the JVM enough heap to hold all your transaction state and query context, and then leave the rest for the page cache. If no page cache memory is configured, then a heuristic setting is computed based on available system resources. By default the size of page cache will be 50% of available RAM minus the max heap size. The size of the page cache will also not be larger than 70x the max heap size (due to some overhead of the page cache in the heap.

A byte size (valid multipliers are B, KiB, KB, K, kB, kb, k, MiB, MB, M, mB, mb, m, GiB, GB, G, gB, gb, g, TiB, TB, PiB, PB, EiB, EB).

Commands to be run when Neo4j Browser successfully connects to this server. Separate multiple commands with semi-colon.

A string.

Whitelist of hosts for the Neo4j Browser to be allowed to fetch content from.

A string.

Configure client applications such as Browser and Bloom to send Product Analytics data.