🚨 Telegraf Error Guide: 'outputs.influxdb database not found' — Fix Missing InfluxDB DB

Overview

The InfluxDB 1.x output plugin logs this when it writes to a database that does not exist and either cannot or is told not to create it:

E! [outputs.influxdb] When writing to [http://influxdb:8086]: database not found: "telemetry"

By default Telegraf tries to create the database on startup. Seeing database not found at write time means creation was skipped, failed silently due to permissions, or the write targets a different name than the one that was created.

Symptoms

• database not found: "..." on every flush; no data ever appears.
• SHOW DATABASES in InfluxDB does not list the name Telegraf is writing to.
• Telegraf was configured with skip_database_creation = true but the database was never provisioned.
• The write user lacks the privilege to create databases, so auto-create quietly failed.

Common Root Causes

• The database simply does not exist and skip_database_creation = true prevented Telegraf from making it.
• Auto-create failed on permissions — the Telegraf user can WRITE but not CREATE DATABASE.
• Name mismatch — Telegraf writes to telemetry but the provisioned database is metrics.
• Wrong server — pointing at a fresh/empty InfluxDB instance that never got provisioned.
• A retention policy referenced in config that does not exist on an otherwise-present database.

Diagnostic Workflow

List the databases that actually exist on the target server:

influx -host influxdb -execute 'SHOW DATABASES'

Compare against the plugin config’s database value:

[[outputs.influxdb]]
  urls = ["http://influxdb:8086"]
  database = "telemetry"
  skip_database_creation = false

Reproduce the write and read the raw error from InfluxDB:

curl -i -XPOST "http://influxdb:8086/write?db=telemetry" \
  --data-binary 'probe,host=diag value=1'

A 404 body of {"error":"database not found: \"telemetry\""} confirms the database is missing. Create it explicitly (with the RP you need):

influx -host influxdb -execute \
  'CREATE DATABASE telemetry WITH DURATION 30d REPLICATION 1 NAME thirty_days'

If you want Telegraf to own creation, ensure the user has the privilege and leave auto-create on:

influx -username admin -password "$ADMIN_PW" \
  -execute 'GRANT ALL ON telemetry TO telegraf'

Example Root Cause Analysis

A Kubernetes deployment set skip_database_creation = true as a “best practice” copied from a blog, assuming an operator would pre-create the database. No such operator step existed, so the fresh InfluxDB had only the internal _internal database. SHOW DATABASES proved telemetry was absent. Because the environment used GitOps, the team added a CREATE DATABASE telemetry bootstrap Job to the same manifest and kept skip_database_creation = true so Telegraf never attempted DDL at scale. On the next apply the database existed before Telegraf’s first flush, and the error cleared.

Prevention Best Practices

• Decide explicitly who owns database creation: either leave skip_database_creation = false with a user that has CREATE DATABASE, or pre-provision it and set skip_database_creation = true — never assume.
• Provision the database (and retention policy) as an idempotent bootstrap step in the same automation that deploys Telegraf.
• Reference the database name via an environment variable so config and provisioning cannot drift.
• With many Telegraf agents, prefer pre-creation plus skip_database_creation = true so hundreds of agents do not each issue DDL.
• Alert on the output error metric so a missing database is caught at deploy, not by an empty dashboard.

Quick Command Reference

# List existing databases
influx -host influxdb -execute 'SHOW DATABASES'

# Reproduce the write
curl -i -XPOST "http://influxdb:8086/write?db=telemetry" \
  --data-binary 'probe value=1'

# Create the database with a retention policy
influx -host influxdb -execute \
  'CREATE DATABASE telemetry WITH DURATION 30d REPLICATION 1'

# Grant the Telegraf user create/write rights
influx -username admin -password "$ADMIN_PW" \
  -execute 'GRANT ALL ON telemetry TO telegraf'

# Watch for the error live
journalctl -u telegraf -f | grep 'database not found'

Conclusion

outputs.influxdb database not found means Telegraf is writing to a database that does not exist, and either skip_database_creation or a missing CREATE DATABASE privilege stopped it from creating one. Confirm with SHOW DATABASES, reconcile the config name, and choose an explicit ownership model for creation. In fleets, pre-provision the database and disable auto-create so agents never issue DDL. For authentication failures instead of missing databases, see the 401 Unauthorized guide. More fixes in the Telegraf guides.