Separate dbt target and state directories to properly support slim CI #69
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jeancochrane
force-pushed
the
jeancochrane/68-data-catalog-dbt-ci-is-not-properly-selecting-for-new-and-modified-models
branch
from
b46d734
to
c1d7d20
Compare
jeancochrane
force-pushed
the
jeancochrane/68-data-catalog-dbt-ci-is-not-properly-selecting-for-new-and-modified-models
branch
from
402d400
to
257ba0f
Compare
…nges" This reverts commit 257ba0f.
jeancochrane
commented
Aug 14, 2023
| @@ -31,15 +31,15 @@ models: | |||
| - pin | |||
| - year | |||
| config: | |||
| error_if: ">280662" | |||
| error_if: ">280679" | |||
There was a problem hiding this comment.
As per usual these days, I also had to bump these thresholds to get tests to pass. Billy is still investigating but his hunch is that the assumptions behind these two tests are not actually correct.
dfsnow
approved these changes
Aug 15, 2023
There was a problem hiding this comment.
This looks great! Nice debugging @jeancochrane. Merge at your leisure.
jeancochrane
deleted the
jeancochrane/68-data-catalog-dbt-ci-is-not-properly-selecting-for-new-and-modified-models
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
One underdocumented subtlety of dbt's state selection options is that when selecting resources based on state attributes (e.g. selecting only resources that are new or modified since last run, for "slim CI") you have to point to a separate state directory containing a manifest file from the previous run, you can't just point to the default state directory that dbt uses (
target/). I had assumed it was best practice to use one state directory, so I designed thebuild-and-test-dbtworkflow to save/restore the state cache from the defaulttarget/state directory; however, it appears that dbt updates its own manifest file before comparing it to the state manifest file to see which resources have changed, so if those manifest files are the same file then dbt will never think that any resources have changed.Another subtlety is that
state:modifiedandstate:neware two separate selectors, so you can't just usestate:modifiedif you want to pick up modified and new models, you have to use both. These two problems are causing the bug described in #68 which is preventing us from deploying changes to our models.This PR fixes our
build-and-test-dbtworkflow such that we compare dbt state to the cached manifest file using a newstate/directory that is separate fromtarget/. It also tweaks the build and test commands to filter forstate:newmodels in addition tostate:modified.Here's an example of a run for a test commit that adds a new model (
default.vw_test) and modifies an existing one (default.vw_pin_universe). You can check the output for theBuild modelsandTest modelssteps to confirm that CI is only building and testing new or modified models.Closes #68.