InfluxDB

Version Differences

• InfluxDB 2.x uses Flux query language, 1.x uses InfluxQL—syntax completely different
• 2.x: buckets, organizations, tokens; 1.x: databases, retention policies, users
• Don't mix documentation—check version before copying queries

Tags vs Fields (Critical)

• Tags are indexed, fields are not—filter on tags, aggregate on fields
• Tag values must be strings—numbers as tags work but waste index space
• Fields support numbers, strings, booleans—store metrics as fields
• Wrong choice kills query performance—can't change after data written

Cardinality Trap

• High-cardinality tags destroy performance—unique user IDs as tags = disaster
• Cardinality = unique combinations of tag values—grows multiplicatively
• Check with SHOW CARDINALITY (1.x) or influx bucket inspect (2.x)
• Rule of thumb: <100K series per measurement; millions = problems