Redshift table size estimates gigabytes as kilobytes.

[vogt4nick]

Describe the bug

The data catalog offers a feature to estimate table size. For Redshift tables, this estimate reports GB as KB.

I can verify this references a table that is 44 GB, not 44 KB.

/~https://github.com/fishtown-analytics/dbt/blob/03010ecde71634d11f80949018c39a63070f1008/plugins/redshift/dbt/include/redshift/macros/catalog.sql#L150

The problem behavior appears to be in this line of code where the bytes are divided by 1e6.

Steps To Reproduce

AFAICT this bug is created by adding a Redshift table to dbt and building the data catalog.

Expected behavior

It should report the correct estimate.

Screenshots and log output

I can verify this references a table that is 44 GB, not 44 KB.

System information

Which database are you using dbt with?

• postgres
• redshift
• bigquery
• snowflake
• other (specify: ____________)

The output of dbt --version:

installed version: 0.17.2
   latest version: 0.17.2

Up to date!

Plugins:
  - postgres: 0.17.2
  - redshift: 0.17.2
  - snowflake: 0.17.2
  - bigquery: 0.17.2

The operating system you're using: Docs are built with Docker image python:3.7.7-slim-buster. Docs are served with Docker image nginx:1.19.0-alpine.

The output of python --version: 3.7.7

Additional context

/~https://github.com/fishtown-analytics/dbt/blob/fb8065df2711d5f5978865a45cc0725a06e670f1/core/dbt/utils.py#L445

Noting because it's related: I'm reasonably confident the issue is not in the format_bytes usage.

[jtcohen6]

Nice catch! Thanks for opening the PR as well. I'll take a look now.

[vogt4nick]

A minor update:

/~https://github.com/fishtown-analytics/dbt/blob/03010ecde71634d11f80949018c39a63070f1008/plugins/redshift/dbt/include/redshift/macros/catalog.sql#L150

The problem behavior appears to be in this line of code where the megabytes are divided by 1e6.

Correction is to multiply by 1e6, not to remove the arithmetic entirely.