@kapa.ai We recently upgraded our Elementary dbt package to Elementary 0.26.0 but upload time grew from ~6.2m to ~8.1m. We're use dbt v1.12 and BigQuery. Any reason for this? Is it regression or some new behaviour?

Thinking completed I'm sorry, but the knowledge sources do not contain specific information about Elementary version 0.26.0 or a documented regression/behavioral change between versions that would explain an increase from ~6.2m to ~8.1m upload time. That said, based on what the knowledge sources do cover, here are the most likely explanations and mitigations: Possible Causes The knowledge sources highlight a known pattern where changes to how

dbt_columns

is updated can impact on-run-end performance in newer versions. [Slack forum] Additionally, the

upload_dbt_columns

macro's

post_hook

walks the full

graph.nodes

and

graph.sources

on every build, and this cost is not gated by standard disable vars. [Linen thread] Other known contributors to upload time growth: • Dynamic Jinja expressions (e.g.,

now()

,

current_timestamp()

) in model/column configs invalidate the metadata hash cache on every run, forcing full re-uploads. [Linen thread] • Project growth — more models/tests/sources increase artifact upload time proportionally. Recommended Mitigations 1. Disable column metadata uploads (biggest impact):

# dbt_project.yml
vars:
  columns_upload_strategy: none

Trade-off:

dbt_columns

won't be updated; column-level schema change monitors and column metadata in Elementary Cloud will show stale data. [Slack forum] 2. Disable artifact auto-upload and run it separately (e.g., only on PR merge or deployment):

vars:
  disable_dbt_artifacts_autoupload: true

Then run manually when needed:

dbt run --select edr.dbt_artifacts

[Linen thread] 3. Check for dynamic expressions in your model configs that may be busting the cache on every run. [Linen thread] Recommendation For a definitive answer on whether this is a regression in 0.26.0 specifically, I'd recommend: • Opening a GitHub issue on the Elementary Data repository referencing the version bump and the timing delta. • Sharing your

dbt.log

with the Elementary team in the #elementary-community Slack channel, as they've used logs to diagnose similar issues in the past. [Slack forum] 👍 0 button 👎 0 button ℹ️ Tip: Mention @elementary.ai in the thread for follow-ups or DM the app for 1-to-1 help.