Skip to content

InfluxDB Operations Guide

This guide covers essential operations for managing InfluxDB, including database creation, data manipulation, and best practices.

Database Management

Create a Database

To initialize a new database, execute the following command:

CREATE DATABASE draas

Retention Policies

Retention policies (RP) define how long data is stored. To set a 3-year retention period as the default for the draas database:

USE draas
CREATE RETENTION POLICY "3years" ON "draas" DURATION 1095d REPLICATION 1 DEFAULT

Data Manipulation

Deleting Data Points

Important

In InfluxDB, you cannot delete a specific field value. You must delete the entire point. Furthermore, you can only use time and tags in the WHERE clause.

How to delete corrupted points :

  1. Query the timestamps first:
SELECT * FROM "DATABASE" WHERE xxxxx > 0 ORDER BY time DESC LIMIT 100
  1. Delete using the timestamp:
DELETE FROM "DATABASE" WHERE time = '2022-07-14T14:55:49.142723763Z'
  1. Delete using Nanoseconds:
# Enter influx with nanosecond precision
influx --precision ns
# Then run the delete
DELETE FROM "matable" WHERE time = 1657809948537456939

Inserting Data

Data is added using the Line Protocol.

-- Format: INSERT <measurement> <fields> <timestamp>
INSERT DATABASE xxxxx=45984 1656637200000000000

Backup and Restore

The portable format is recommended for all backup operations.

  • Backup: influxd backup -portable -db pdu /root/influxbkp
  • Restore: influxd restore -portable -db pdu /root/influxbkp

Core Concepts

  • Measurement: Similar to a SQL Table.
  • Fields: The actual data (e.g., temperature). Not indexed.
  • Tags: Key-value metadata. Indexed.
  • Point: A single record (Measurement + Tags + Fields + Time).
  • Series: A collection of points sharing the same measurement and tags.

Database Design Best Practices

Optimize Series Cardinality

Series cardinality is the total number of unique tag combinations. High cardinality causes high memory usage and crashes.

  • Do not use tags for highly variable data (like random IDs, specific sensor values, or timestamps).
  • Do use tags for data you frequently filter or group by.