Config Options

{{< product-name >}} lets you customize your configuration by using influxdb3 command options or by setting environment variables.

Configure your server

Pass configuration options using either command options or environment variables. Command options take precedence over environment variables.

Global vs serve-specific options

Some options are global (specified before the command) while others are serve-specific (specified after serve):

• Global options: Apply to the influxdb3 CLI itself (for example, --num-io-threads)
• Serve options: Apply only to the serve command (for example, --node-id, --object-store, --verbose)

Example command with global and serve-specific options

{{% show-in "core" %}}

influxdb3 --num-io-threads=4 serve \
  --node-id node0 \
  --object-store file \
  --data-dir ~/.influxdb3 \
  --verbose \
  --log-filter info

{{% /show-in %}}

{{% show-in "enterprise" %}}

influxdb3 --num-io-threads=4 serve \
  --node-id node0 \
  --cluster-id cluster0 \
  --license-email [email protected] \
  --object-store file \
  --data-dir ~/.influxdb3 \
  --verbose \
  --log-filter info

{{% /show-in %}}

Example environment variables

{{% show-in "enterprise" %}}export [email protected]
export INFLUXDB3_ENTERPRISE_CLUSTER_ID=cluster0
{{% /show-in %}}export INFLUXDB3_NODE_IDENTIFIER_PREFIX=my-node
export INFLUXDB3_OBJECT_STORE=file
export INFLUXDB3_DB_DIR=~/.influxdb3
export INFLUXDB3_LOG_FILTER=info

influxdb3 serve

Global configuration options

The following options apply to the influxdb3 CLI globally and must be specified before any subcommand (for example, serve):

num-io-threads

Sets the number of threads allocated to the IO runtime thread pool. IO threads handle HTTP request serving, line protocol parsing, and file operations.

[!Important] --num-io-threads is a global option that must be specified before the serve command.

{{% show-in "enterprise" %}} Default: 2 {{% /show-in %}}

# Set IO threads (global option before serve)
influxdb3 --num-io-threads=8 serve --node-id=node0 --object-store=file

{{% show-in "enterprise" %}} For detailed information about thread allocation, see the Resource Limits section. {{% /show-in %}}

Server configuration options

• General {{% show-in "enterprise" %}}- Licensing{{% /show-in %}}
• Security
• AWS
• Google Cloud Service
• Microsoft Azure
• Object Storage
• Logs
• Traces
• DataFusion
• HTTP
• Memory
• Write-Ahead Log (WAL)
• Compaction
• Caching
• Processing Engine {{% show-in "enterprise" %}}
• Cluster Management {{% /show-in %}}
• Resource Limits
• Data Lifecycle Management
• Telemetry
• TCP Listeners

General

{{% show-in "enterprise" %}}

• cluster-id {{% /show-in %}}
• data-dir {{% show-in "enterprise" %}}
• mode {{% /show-in %}}
• node-id {{% show-in "enterprise" %}}
• node-id-from-env {{% /show-in %}}
• object-store
• query-file-limit {{% show-in "enterprise" %}}
• use-pacha-tree {{% /show-in %}}

{{% show-in "enterprise" %}}

cluster-id

Specifies the cluster identifier that prefixes the object store path for the Enterprise Catalog. This value must be different than the --node-id value.

{{% /show-in %}}

data-dir

For the file object store, defines the location {{< product-name >}} uses to store files locally. Required when using the file object store.

{{% show-in "enterprise" %}}

mode

Sets the mode to start the server in, allowing you to create specialized nodes in a distributed cluster.

This option supports the following values:

• all (default): Enables all server modes
• ingest: Enables only data ingest capabilities
• query: Enables only query capabilities
• compact: Enables only compaction processes
• process: Enables only data processing capabilities

You can specify multiple modes using a comma-delimited list (for example, ingest,query).

Default: all

[!Important] Thread allocation for different modes:

• Ingest mode: Benefits from additional IO threads for line protocol parsing. For high-throughput scenarios with multiple concurrent writers, consider increasing --num-io-threads (global option) to 8-16+ to optimize performance. DataFusion threads are still needed for snapshot operations.

• Query mode: Benefits from maximizing DataFusion threads. Use most available cores for DataFusion with minimal IO threads (2-4).

• Compact mode: Primarily uses DataFusion threads for sort/dedupe operations.

• All mode: Requires balanced thread allocation based on your workload mix.

Example configurations:

# High-throughput ingest node (32 cores)
influxdb3 --num-io-threads=12 serve --mode=ingest --datafusion-num-threads=20

# Query-optimized node (32 cores)
influxdb3 --num-io-threads=4 serve --mode=query --datafusion-num-threads=28

# Balanced all-in-one (32 cores)
influxdb3 --num-io-threads=6 serve --mode=all --datafusion-num-threads=26

{{% /show-in %}}

node-id

Specifies the node identifier used as a prefix in all object store file paths. This should be unique for any hosts sharing the same object store configuration--for example, the same bucket.

{{% show-in "enterprise" %}}

node-id-from-env

Specifies the node identifier used as a prefix in all object store file paths. Takes the name of an environment variable as an argument and uses the value of that environment variable as the node identifier. This option cannot be used with the --node-id option.

Example using --node-id-from-env

export DATABASE_NODE=node0 && influxdb3 serve \
  --node-id-from-env DATABASE_NODE \
  --cluster-id cluster0 \
  --object-store file \
  --data-dir ~/.influxdb3/data

{{% /show-in %}}

object-store

Specifies which object storage to use to store Parquet files. This option supports the following values:

• memory: Effectively no object persistence
• memory-throttled: Like memory but with latency and throughput that somewhat resembles a cloud object store
• file: Stores objects in the local filesystem (must also set --data-dir)
• s3: Amazon S3 (must also set --bucket, --aws-access-key-id, --aws-secret-access-key, and possibly --aws-default-region)
• google: Google Cloud Storage (must also set --bucket and --google-service-account)
• azure: Microsoft Azure blob storage (must also set --bucket, --azure-storage-account, and --azure-storage-access-key)

{{% show-in "enterprise" %}}

use-pacha-tree <span class="badge experimental">Experimental</span> {#use-pacha-tree}

Enables the PachaTree storage engine.

[!Caution] PachaTree is an experimental feature not for production use. It might not be compatible with other features and configuration options.

Default: false

{{% /show-in %}}

{{% show-in "enterprise" %}}

Licensing

license-email

Specifies the email address to associate with your {{< product-name >}} license and automatically responds to the interactive email prompt when the server starts. This option is mutually exclusive with license-file.

license-file

Specifies the path to a license file for {{< product-name >}}. When provided, the license file's contents are used instead of requesting a new license. This option is mutually exclusive with license-email.

license-type

Specifies the type of {{% product-name %}} license to use and bypasses the interactive license prompt. Provide one of the following license types:

• home
• trial
• commercial

{{% /show-in %}}

Security

• tls-key
• tls-cert
• tls-minimum-versions
• without-auth
• disable-authz
• admin-token-recovery-http-bind
• admin-token-file {{% show-in "enterprise" %}}- permission-tokens-file{{% /show-in %}}

tls-key

The path to a key file for TLS to be enabled.

tls-cert

The path to a cert file for TLS to be enabled.

tls-minimum-version

The minimum version for TLS. Valid values are tls-1.2 or tls-1.3. Default is tls-1.2.

without-auth

Disables authentication for all server actions (CLI commands and API requests). The server processes all requests without requiring tokens or authentication.

disable-authz

Optionally disable authz by passing in a comma separated list of resources. Valid values are health, ping, and metrics.

admin-token-recovery-http-bind

Enables an admin token recovery HTTP server on a separate port. This server allows regenerating lost admin tokens without existing authentication. The server automatically shuts down after a successful token regeneration.

[!Warning] This option creates an unauthenticated endpoint that can regenerate admin tokens. Only use this when you have lost access to your admin token and ensure the server is only accessible from trusted networks.

Default: 127.0.0.1:8182 (when enabled)

Example usage

# Start server with recovery endpoint
influxdb3 serve --admin-token-recovery-http-bind

# In another terminal, regenerate the admin token
influxdb3 create token --admin --regenerate --host http://127.0.0.1:8182

admin-token-file

Specifies an offline admin token file to use if no tokens exist when the server starts. Once started, you can interact with the server using the provided token. Offline admin tokens are designed to help with automated deployments.

Offline admin tokens are defined in a JSON-formatted file. Use the following command to generate an offline admin token file:

influxdb3 create token --admin \
  --name "example-admin-token" \
  --expiry 1d \
  --offline \
  --output-file ./path/to/admin-token.json

{{< expand-wrapper >}} {{% expand "View example offline admin token file" %}}

{
  "token": "apiv3_0XXXX-xxxXxXxxxXX_OxxxX...",
  "name": "example-admin-token",
  "expiry_millis": 1756400061529
}

{{% /expand %}} {{< /expand-wrapper >}}

Example usage

# Generate an admin token offline
influxdb3 create token \
  --admin \
  --name "example-admin-token" \
  --expiry 1d \
  --offline \
  --output-file ./path/to/admin-token.json

# Start {{% product-name %}} using the generated token
influxdb3 serve --admin-token-file ./path/to/admin-token.json

{{% show-in "enterprise" %}}

permission-tokens-file

Specifies an offline permission (resource) tokens file to use if no resource tokens exist when the server starts. Once started, you can interact with the server using the provided tokens. Offline permission tokens are designed to help with automated deployments.

Multiple tokens with database-level permissions can be defined. You can also specify databases to create at startup. Use the a command similar to the following to generate an offline permission token file:

influxdb3 create token \
  --name "example-token" \
  --permission "db:db1,db2:read,write" \
  --permission "db:db3:read" \
  --expiry 1d \
  --offline \
  --create-databases db1,db2 \
  --output-file ./path/to/tokens.json

{{< expand-wrapper >}} {{% expand "View example offline permission tokens file" %}}

{
  "create_databases": [
    "db1",
    "db2",
    "d3"
  ],
  "tokens": [
    {
      "token": "apiv3_0XXXX-xxxXxXxxxXX_OxxxX...",
      "name": "example-token",
      "expiry_millis": 1756400061529,
      "permissions": [
        "db:db1,db2:read,write",
        "db:db3:read"
      ]
    }
  ]
}

{{% /expand %}} {{< /expand-wrapper >}}

[!Note] If you write a new offline permission token to an existing permission token file, the command appends the new token to the existing output file.

Example usage

# Generate an admin token offline
influxdb3 create token \
  --name "example-token" \
  --permission "db:db1,db2:read,write" \
  --permission "db:db3:read" \
  --expiry 1d \
  --offline \
  --create-databases db1,db2 \
  --output-file ./path/to/tokens.json

# Start {{% product-name %}} using the generated token
influxdb3 serve --permission-tokens-file ./path/to/tokens.json

{{% /show-in %}}

AWS

• aws-access-key-id
• aws-secret-access-key
• aws-default-region
• aws-endpoint
• aws-session-token
• aws-allow-http
• aws-skip-signature
• aws-credentials-file

aws-access-key-id

When using Amazon S3 as the object store, set this to an access key that has permission to read from and write to the specified S3 bucket.

aws-secret-access-key

When using Amazon S3 as the object store, set this to the secret access key that goes with the specified access key ID.

aws-default-region

When using Amazon S3 as the object store, set this to the region that goes with the specified bucket if different from the fallback value.

Default: us-east-1

aws-endpoint

When using an Amazon S3 compatibility storage service, set this to the endpoint.

aws-session-token

When using Amazon S3 as an object store, set this to the session token. This is handy when using a federated login or SSO and fetching credentials via the UI.

aws-allow-http

Allows unencrypted HTTP connections to AWS.

aws-skip-signature

If enabled, S3 object stores do not fetch credentials and do not sign requests.

aws-credentials-file

Specifies the path to your S3 credentials file. When using a credentials file, settings in the file override the corresponding CLI flags.

S3 credential files are JSON-formatted and should contain the following:

{
  "aws_access_key_id": "AWS_ACCESS_KEY_ID",
  "aws_secret_access_key": "AWS_SECRET_ACCESS_KEY",
  "aws_session_token": "AWS_SESSION_TOKEN",
  "expiry": "UNIX_SECONDS_TIMESTAMP"
}

The aws_session_token and expiry fields are optional. The file is automatically checked for updates at the expiry time or at 1-hour intervals.

If the object store returns an "Unauthenticated" error, InfluxDB will attempt to update its in-memory credentials from this file and then retry the object store request.

Google Cloud Service

• google-service-account

google-service-account

When using Google Cloud Storage as the object store, set this to the path to the JSON file that contains the Google credentials.

Microsoft Azure

• azure-storage-account
• azure-storage-access-key
• azure-endpoint
• azure-allow-http

azure-storage-account

When using Microsoft Azure as the object store, set this to the name you see when navigating to All Services > Storage accounts > [name].

azure-storage-access-key

When using Microsoft Azure as the object store, set this to one of the Key values in the Storage account's Settings > Access keys.

azure-endpoint

When using Microsoft Azure as the object store, set this to the Azure Blob Storage endpoint.

azure-allow-http

When using Microsoft Azure as the object store, allow unencrypted HTTP requests to Azure Blob Storage.

Default: false

Object Storage

• bucket
• object-store-connection-limit
• object-store-http2-only
• object-store-http2-max-frame-size
• object-store-max-retries
• object-store-retry-timeout
• object-store-cache-endpoint

bucket

Sets the name of the object storage bucket to use. Must also set --object-store to a cloud object storage for this option to take effect.

object-store-connection-limit

When using a network-based object store, limits the number of connections to this value.

Default: 16

object-store-http2-only

Forces HTTP/2 connections to network-based object stores.

object-store-http2-max-frame-size

Sets the maximum frame size (in bytes/octets) for HTTP/2 connections.

object-store-max-retries

Defines the maximum number of times to retry a request.

object-store-retry-timeout

Specifies the maximum length of time from the initial request after which no further retries are be attempted.

object-store-cache-endpoint

Sets the endpoint of an S3-compatible, HTTP/2-enabled object store cache.

Logs

• log-filter
• log-destination
• log-format
• query-log-size

log-filter

Sets the filter directive for logs. Use this option to control the verbosity of server logs globally or for specific components.

Log levels

The following log levels are available (from least to most verbose):

Basic usage

To set the log level globally, pass one of the log levels:

influxdb3 serve --log-filter debug

Targeted filtering

Globally enabling debug or trace produces a high volume of log output. For more targeted debugging, you can set different log levels for specific components using the format <global_level>,<component>=<level>.

Debug write buffer operations

influxdb3 serve --log-filter info,influxdb3_write_buffer=debug

Trace WAL operations

influxdb3 serve --log-filter info,influxdb3_wal=trace

Multiple targeted filters

influxdb3 serve --log-filter info,influxdb3_write_buffer=debug,influxdb3_wal=debug

{{% show-in "enterprise" %}}

Debug Enterprise storage engine operations

influxdb3 serve --log-filter info,influxdb3_enterprise=debug

{{% /show-in %}}

Common component names

The following are common component names you can use for targeted filtering:

[!Note] Targeted filtering requires knowledge of the codebase component names. The component names correspond to Rust package names in the InfluxDB 3 source code. Use debug or trace sparingly on specific components to avoid excessive log output.

log-destination

Specifies the destination for logs.

This option supports the following values:

• stdout (default)
• stderr

Default: stdout

log-format

Defines the message format for logs.

This option supports the following values:

• full (default)

Default: full

query-log-size

Defines the size of the query log. Up to this many queries remain in the log before older queries are evicted to make room for new ones.

Default: 1000

Traces

• traces-exporter
• traces-exporter-jaeger-agent-host
• traces-exporter-jaeger-agent-port
• traces-exporter-jaeger-service-name
• traces-exporter-jaeger-trace-context-header-name
• traces-jaeger-debug-name
• traces-jaeger-tags
• traces-jaeger-max-msgs-per-second

traces-exporter

Sets the type of tracing exporter.

Default: none

traces-exporter-jaeger-agent-host

Specifies the Jaeger agent network hostname for tracing.

Default: 0.0.0.0

traces-exporter-jaeger-agent-port

Defines the Jaeger agent network port for tracing.

Default: 6831

traces-exporter-jaeger-service-name

Sets the Jaeger service name for tracing.

Default: iox-conductor

traces-exporter-jaeger-trace-context-header-name

Specifies the header name used for passing trace context.

Default: uber-trace-id

traces-jaeger-debug-name

Specifies the header name used for force sampling in tracing.

Default: jaeger-debug-id

traces-jaeger-tags

Defines a set of key=value pairs to annotate tracing spans with.

traces-jaeger-max-msgs-per-second

Specifies the maximum number of messages sent to a Jaeger service per second.

Default: 1000

DataFusion

• datafusion-num-threads
• datafusion-max-parquet-fanout
• datafusion-use-cached-parquet-loader
• datafusion-config

datafusion-num-threads

Sets the maximum number of DataFusion runtime threads to use.

datafusion-max-parquet-fanout

When multiple Parquet files are required in a sorted way (deduplication for example), specifies the maximum fanout.

Default: 1000

datafusion-use-cached-parquet-loader

Uses a cached Parquet loader when reading Parquet files from the object store.

datafusion-config

Provides custom configuration to DataFusion as a comma-separated list of key:value pairs.

HTTP

• max-http-request-size
• http-bind

max-http-request-size

Specifies the maximum size of HTTP requests.

Default: 10485760

http-bind

Defines the address on which InfluxDB serves HTTP API requests.

Default: 0.0.0.0:8181

Memory

• exec-mem-pool-bytes
• force-snapshot-mem-threshold

exec-mem-pool-bytes

Specifies the size of the memory pool used for query processing and data operations. This memory pool is used when {{% product-name %}} processes queries and performs internal data management tasks. Can be given as absolute value in bytes or as a percentage of the total available memory--for example: 8000000000 or 10%.

{{% show-in "core" %}}Default: 8589934592{{% /show-in %}} {{% show-in "enterprise" %}}Default: 20%{{% /show-in %}}

force-snapshot-mem-threshold

Specifies the threshold for the internal memory buffer. Supports either a percentage (portion of available memory) or absolute value in MB--for example: 70% or 1000.

{{% show-in "core" %}}Default: 70%{{% /show-in %}} {{% show-in "enterprise" %}}Default: 50%{{% /show-in %}}

Write-Ahead Log (WAL)

• checkpoint-interval
• wal-flush-interval
• wal-snapshot-size
• wal-max-write-buffer-size
• snapshotted-wal-files-to-keep
• wal-replay-fail-on-error
• wal-replay-concurrency-limit

checkpoint-interval {#checkpoint-interval metadata="v3.8.2+"}

Sets the interval for consolidating snapshots into monthly checkpoints for faster server startup. Snapshots accumulate in object storage over time and are not automatically deleted.

Without checkpointing, the server loads individual snapshots on startup. The number of snapshots is determined by the lookback window (gen1-lookback-duration, default 1 month) divided by gen1-duration (default 10 minutes), with a minimum of 100. With default settings, that can be up to ~4,320 snapshots.

With checkpointing enabled, the server periodically consolidates snapshot metadata into checkpoints in object storage. On startup, the server loads one to two checkpoints per calendar month, then loads only snapshots created since the last checkpoint. Enabling checkpointing does not delete old snapshots.

Up to 10 checkpoints load concurrently during startup. The server retains two checkpoints per calendar month and handles month rollovers automatically.

Accepts a duration value--for example: 1h, 30m, 10m.

Default: Not set (disabled)

Example

influxdb3 serve --checkpoint-interval 1h

wal-flush-interval

Specifies the interval to flush buffered data to a WAL file. Writes that wait for WAL confirmation take up to this interval to complete. Use s for seconds or ms for milliseconds. For local disks, 100 ms is recommended.

Default: 1s

wal-snapshot-size

Defines the number of WAL files to attempt to remove in a snapshot. This, multiplied by the interval, determines how often snapshots are taken.

Default: 600

wal-max-write-buffer-size

Specifies the maximum number of write requests that can be buffered before a flush must be executed and succeed.

Default: 100000

snapshotted-wal-files-to-keep

Specifies the number of snapshotted WAL files to retain in the object store. Flushing the WAL files does not clear the WAL files immediately; they are deleted when the number of snapshotted WAL files exceeds this number.

Default: 300

wal-replay-fail-on-error

Determines whether WAL replay should fail when encountering errors.

Default: false

wal-replay-concurrency-limit

Concurrency limit during WAL replay. Setting this number too high can lead to OOM. The default is dynamically determined.

Default: max(num_cpus, 10)

Compaction

{{% show-in "enterprise" %}}

• compaction-max-num-files-per-plan
• compaction-gen2-duration
• compaction-multipliers
• compaction-cleanup-wait
• compaction-check-interval {{% /show-in %}}
• gen1-duration

{{% show-in "enterprise" %}}

compaction-max-num-files-per-plan

Sets the maximum number of files included in any compaction plan.

Default: 500

compaction-gen2-duration

Specifies the duration of the first level of compaction (gen2). Later levels of compaction are multiples of this duration. This value should be equal to or greater than the gen1 duration.

Default: 20m

compaction-multipliers

Specifies a comma-separated list of multiples defining the duration of each level of compaction. The number of elements in the list determines the number of compaction levels. The first element specifies the duration of the first level (gen3); subsequent levels are multiples of the previous level.

Default: 3,4,6,5

compaction-cleanup-wait

Specifies the amount of time that the compactor waits after finishing a compaction run to delete files marked as needing deletion during that compaction run.

Default: 10m

compaction-check-interval

Specifies how often the compactor checks for new compaction work to perform.

Default: 10s

{{% /show-in %}}

gen1-duration

Specifies the duration that Parquet files are arranged into. Data timestamps land each row into a file of this duration. Supported durations are 1m, 5m, and 10m. These files are known as "generation 1" files{{% show-in "enterprise" %}}, which the compactor can merge into larger generations{{% /show-in %}}{{% show-in "core" %}} that the compactor in InfluxDB 3 Enterprise can merge into larger generations{{% /show-in %}}.

Default: 10m

Caching

• preemptive-cache-age
• parquet-mem-cache-size
• parquet-mem-cache-prune-percentage
• parquet-mem-cache-prune-interval
• parquet-mem-cache-query-path-duration
• disable-parquet-mem-cache
• table-index-cache-max-entries
• table-index-cache-concurrency-limit {{% show-in "enterprise" %}}
• last-value-cache-disable-from-history {{% /show-in %}}
• last-cache-eviction-interval {{% show-in "enterprise" %}}
• distinct-value-cache-disable-from-history {{% /show-in %}}
• distinct-cache-eviction-interval

preemptive-cache-age

Specifies the interval to prefetch into the Parquet cache during compaction.

Default: 3d

parquet-mem-cache-size

Specifies the size of the in-memory Parquet cache. Accepts values in megabytes (as an integer) or as a percentage of total available memory (for example, 20%, 4096).

Default: 20%

[!Note]

Breaking change in v3.0.0

In v3.0.0, --parquet-mem-cache-size-mb was replaced with --parquet-mem-cache-size. The new option accepts both megabytes (integer) and percentage values. The default changed from 1000 MB to 20% of total available memory.

parquet-mem-cache-prune-percentage

Specifies the percentage of entries to prune during a prune operation on the in-memory Parquet cache.

Default: 0.1

parquet-mem-cache-prune-interval

Sets the interval to check if the in-memory Parquet cache needs to be pruned.

Default: 1s

parquet-mem-cache-query-path-duration

{{% show-in "enterprise" %}} A duration that specifies {{% /show-in %}}{{% show-in "core" %}} Specifies {{% /show-in %}} the time window for caching recent Parquet files in memory. Default is 5h.

Only files containing data with a timestamp between now and now - duration are cached when accessed during queries--for example, with the default 5h setting:

• Current time: 2024-06-10 15:00:00
• Cache window: Last 5 hours (2024-06-10 10:00:00 to now)

If a query requests data from 2024-06-09 (old) and 2024-06-10 14:00 (recent):

• Cached: Parquet files with data from 2024-06-10 14:00 (within 5-hour window)
• Not cached: Parquet files with data from 2024-06-09 (outside 5-hour window)

disable-parquet-mem-cache

Disables the in-memory Parquet cache. By default, the cache is enabled.

table-index-cache-max-entries

Specifies the maximum number of entries in the table index cache.

Default: 1000

table-index-cache-concurrency-limit

Limits the concurrency level for table index cache operations.

Default: 8

{{% show-in "enterprise" %}}

last-value-cache-disable-from-history

Disables populating the last-N-value cache from historical data. If disabled, the cache is still populated with data from the write-ahead log (WAL).

{{% /show-in %}}

last-cache-eviction-interval

Specifies the interval to evict expired entries from the Last-N-Value cache, expressed as a human-readable duration--for example: 20s, 1m, 1h.

Default: 10s

{{% show-in "enterprise" %}}

distinct-value-cache-disable-from-history

Disables populating the distinct value cache from historical data. If disabled, the cache is still populated with data from the write-ahead log (WAL).

{{% /show-in %}}

distinct-cache-eviction-interval

Specifies the interval to evict expired entries from the distinct value cache, expressed as a human-readable duration--for example: 20s, 1m, 1h.

Default: 10s

query-file-limit

Limits the number of Parquet files a query can access. If a query attempts to read more than this limit, {{< product-name >}} returns an error.

{{% show-in "core" %}} Default: 432

With the default 432 setting and the default gen1-duration setting of 10 minutes, queries can access up to a 72 hours of data, but potentially less depending on whether all data for a given 10 minute block of time was ingested during the same period.

You can increase this limit to allow more files to be queried, but be aware of the following side-effects:

• Degraded query performance for queries that read more Parquet files
• Increased memory usage
• Your system potentially killing the influxdb3 process due to Out-of-Memory (OOM) errors
• If using object storage to store data, many GET requests to access the data (as many as 2 per file)

[!Note] We recommend keeping the default setting and querying smaller time ranges. If you need to query longer time ranges or faster query performance on any query that accesses an hour or more of data, InfluxDB 3 Enterprise optimizes data storage by compacting and rearranging Parquet files to achieve faster query performance. {{% /show-in %}}

Processing Engine

• plugin-dir
• plugin-repo
• virtual-env-location
• package-manager

plugin-dir

Specifies the local directory that contains Python plugins and their test files.

Default behavior by deployment type

Disable the Processing Engine

To disable the Processing Engine, ensure plugin-dir is not configured.

[!Warning] Setting plugin-dir="" or INFLUXDB3_PLUGIN_DIR="" (empty string) does not disable the Processing Engine. You must comment out, remove, or unset the configuration — not set it to empty.

{{% show-in "enterprise" %}} Docker: Use INFLUXDB3_UNSET_VARS to unset default environment variables that are preconfigured in the container image.

INFLUXDB3_UNSET_VARS accepts a comma-separated list of environment variable names to unset in the container entrypoint before {{< product-name >}} starts. This lets you disable or override image defaults (for example, INFLUXDB3_PLUGIN_DIR, logging, or other configuration variables) without modifying the container image itself.

To disable the default plugin directory, unset INFLUXDB3_PLUGIN_DIR:

docker run -e INFLUXDB3_UNSET_VARS="INFLUXDB3_PLUGIN_DIR" influxdb:3-enterprise

{{% /show-in %}}

{{% show-in "core" %}} Docker: Use a custom entrypoint:

docker run --entrypoint /bin/sh influxdb:3-core -c 'unset INFLUXDB3_PLUGIN_DIR && exec influxdb3 serve --object-store memory'

{{% /show-in %}}

systemd (DEB/RPM): Comment out or remove plugin-dir in the configuration file:

sudo nano /etc/influxdb3/influxdb3-{{< product-key >}}.conf

# plugin-dir="/var/lib/influxdb3/plugins"

Then restart the service:

sudo systemctl restart influxdb3-{{< product-key >}}

When the Processing Engine is disabled:

• The Python environment and PyO3 bindings are not initialized
• Plugin-related operations return a "No plugin directory configured" error
• The server runs with reduced resource usage

plugin-repo

Specifies the base URL of the remote repository used when referencing plugins with the gh: prefix. When you create a trigger with a plugin filename starting with gh:, InfluxDB fetches the plugin code from this repository URL.

The URL construction automatically handles trailing slashes—both formats work identically:

• https://example.com/plugins/ (with trailing slash)
• https://example.com/plugins (without trailing slash)

Default: The official InfluxDB 3 plugins repository at https://raw.githubusercontent.com/influxdata/influxdb3_plugins/main/

Example usage

# Use a custom organization repository
influxdb3 serve \
  --plugin-dir ~/.plugins \
  --plugin-repo "https://raw.githubusercontent.com/myorg/influxdb-plugins/main/"

# Use an internal mirror
influxdb3 serve \
  --plugin-dir ~/.plugins \
  --plugin-repo "https://internal.company.com/influxdb-plugins/"

# Set via environment variable
export INFLUXDB3_PLUGIN_REPO="https://custom-repo.example.com/plugins/"
influxdb3 serve --plugin-dir ~/.plugins

virtual-env-location

Specifies the location of the Python virtual environment that the processing engine uses.

package-manager

Specifies the Python package manager that the Processing Engine uses to install plugin dependencies.

This option supports the following values:

• discover (default): Automatically detect and use available package manager (uv or pip)
• pip: Use pip package manager exclusively
• uv: Use uv package manager exclusively
• disabled: Disable automatic package installation (all dependencies must be pre-installed)

Default: discover

Security mode (disabled)

When set to disabled, {{< product-name >}} blocks all package installation attempts for security and compliance requirements. The Processing Engine and plugins continue to function normally, but package installation requests (via CLI or API) return a 403 Forbidden error.

This mode is designed for:

• Enterprise security requirements: Prevent arbitrary package installation
• Compliance environments: Control exactly which packages are available
• Air-gapped deployments: Pre-install all dependencies before deployment
• Multi-tenant scenarios: Prevent tenants from installing potentially malicious packages

[!Important] Before using --package-manager disabled, administrators must pre-install all required Python packages into the virtual environment that plugins will use.

Example:

# Start InfluxDB 3 with disabled package manager
influxdb3 serve \
  --node-id node0 \
  --object-store file \
  --data-dir ~/.influxdb3 \
  --plugin-dir ~/.plugins \
  --package-manager disabled

For more information about plugins and package management, see Processing Engine plugins.

{{% show-in "enterprise" %}}

Cluster Management

• replication-interval
• catalog-sync-interval
• wait-for-running-ingestor

replication-interval

Specifies the interval at which data replication occurs between cluster nodes.

Default: 250ms

catalog-sync-interval

Defines how often the catalog synchronizes across cluster nodes.

Default: 10s

wait-for-running-ingestor

Specifies how long to wait for a running ingestor during startup.

Default: 10s

{{% /show-in %}}

Resource Limits

{{% show-in "enterprise" %}}

• num-cores {{% /show-in %}}
• datafusion-num-threads
• num-io-threads - See Global configuration options {{% show-in "enterprise" %}}
• num-database-limit
• num-table-limit
• num-total-columns-per-table-limit

num-cores

Limits the number of CPU cores that the InfluxDB 3 Enterprise process can use when running on systems where resources are shared.

Default: All available cores on the system

Maximum cores allowed is determined by your {{% product-name %}} license:

• Trial: up to 256 cores
• At-Home: 2 cores
• Commercial: per contract

When specified, InfluxDB automatically assigns the number of DataFusion threads and IO threads based on the core count.

Default thread assignment logic when num-cores is set:

• 1-2 cores: 1 IO thread, 1 DataFusion thread
• 3 cores: 1 IO thread, 2 DataFusion threads
• 4+ cores: 2 IO threads, (n-2) DataFusion threads

This automatic allocation applies when you don't explicitly set --num-io-threads and --datafusion-num-threads.

[!Note] You can override the automatic thread assignment by explicitly setting --num-io-threads and --datafusion-num-threads. This is particularly important for specialized workloads like ingest mode where you may need more IO threads than the default allocation.

Constraints:

• Must be at least 2
• Cannot exceed the number of cores available on the system
• Total thread count from --num-io-threads (global option) and --datafusion-num-threads cannot exceed the num-cores value

datafusion-num-threads

Sets the number of threads allocated to the DataFusion runtime thread pool. DataFusion threads handle:

• Query execution and processing
• Data aggregation and transformation
• Snapshot creation (sort/dedupe operations)
• Parquet file generation

{{% show-in "core" %}} Default: All available cores minus IO threads

[!Note] DataFusion threads are used for both query processing and snapshot operations. {{% /show-in %}}

{{% show-in "enterprise" %}} Default:

• If not specified and --num-cores is not set: All available cores minus IO threads
• If not specified and --num-cores is set: Automatically determined based on core count (see --num-cores)

[!Note] DataFusion threads are used for both query processing and snapshot operations. Even ingest-only nodes use DataFusion threads during WAL snapshot creation.

Constraints: When used with --num-cores, the sum of --num-io-threads and --datafusion-num-threads cannot exceed the num-cores value {{% /show-in %}}

[!Note] --num-io-threads is a global configuration option.

{{% show-in "enterprise" %}}

num-database-limit

Limits the total number of active databases. Default is {{% influxdb3/limit "database" %}}.

num-table-limit

Limits the total number of active tables across all databases. Default is {{% influxdb3/limit "table" %}}.

num-total-columns-per-table-limit

Limits the total number of columns per table. Default is {{% influxdb3/limit "column" %}}.

{{% /show-in %}}

Data Lifecycle Management

• gen1-lookback-duration
• retention-check-interval
• delete-grace-period
• hard-delete-default-duration

gen1-lookback-duration

Specifies how far back to look when creating generation 1 Parquet files.

Default: 24h

retention-check-interval

The interval at which retention policies are checked and enforced. Enter as a human-readable time--for example: 30m or 1h.

Default: 30m

delete-grace-period

Specifies the grace period before permanently deleting data.

Default: 24h

hard-delete-default-duration

Sets the default duration for hard deletion of data.

Default: 90d

Telemetry

• disable-telemetry-upload
• telemetry-endpoint

disable-telemetry-upload

Disables the upload of telemetry data to InfluxData.

Default: false

telemetry-endpoint

Specifies the endpoint for telemetry data uploads.

TCP Listeners

• tcp-listener-file-path
• admin-token-recovery-tcp-listener-file-path

tcp-listener-file-path

Specifies the file path for the TCP listener configuration.

admin-token-recovery-tcp-listener-file-path

Specifies the TCP listener file path for admin token recovery operations.