> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mage.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# InfluxDB

## InfluxDB instance

InfluxDB can be [run locally](https://docs.influxdata.com/influxdb/v2.7/install/) or in the cloud.

## Basic config

Basic configuration includes the server connection details:

```yaml theme={"system"}
connector_type: influxdb
url: "http://localhost:8086"
token: "my-token"
org: "my-org"
bucket: "data"
```

## Data format

InfluxDB supports [data elements](https://docs.influxdata.com/influxdb/v2.7/reference/key-concepts/data-elements/) for structuring your data.
The influxdb data exporter can create these elements if the messages received contain a metadata dictionary:

```python theme={"system"}
message['metadata'] = {
    'measurement': str
    'time': int
    'tags': Dict
}
```

The following message shows the configuration of all possible elements data and metadata:

```python theme={"system"}
message = {
    "data": {'bees': 23},
    "metadata": {
        'measurement': 'census',
        'time': 1693472675783,  # timestamp with ms precision
        'tags': {'location': 'klamath', 'scientist': 'anderson'}
    }
}
```

## Advanced config

The influxdb data reader reads data from InfluxDB server in timed intervals.
Every `timeout_ms` data with time stamps in the interval `(t-time_delay, t-time_delay+timeout_ms)` will be returned, where `t` is the current time.
Note that InfluxDB is not a streaming database, so trying to set `time_delay` too small may result in data loss.
This will occur if data is written to the database after the corresponding interval has been read.
The time delay is given as [InfluxDB duration](https://docs.influxdata.com/flux/v0.x/data-types/basic/duration/).

The `batch_size` limits the size of batches returned by the data reader.
If `print_intervals` is `true` the intervals are printed on every time step.
The `filter_fn` can be used to further specify the data, e.g. `r._measurement == "cpu" and r._field == "usage_system"` (note the usage of single quotation marks outside and double quotation marks inside the filter\_fn string).
Defaults to `true` for no further filtering.
For documentation, see [InfluxDB filter() function](https://docs.influxdata.com/flux/v0.x/stdlib/universe/filter/)

```yaml theme={"system"}
connector_type: influxdb
url: "http://localhost:8086"
token: "my-token"
org: "my-org"
bucket: "data"
time_delay: "5m"
timeout_ms: 1000
batch_size: 100
print_intervals: false
filter_fn: 'true'
```