https://linen.dev logo
Join Slack
Powered by
<@U088WCP3B4P> How do I configure a custom schema ...
# ask-elementary-ai
v
@kapa.ai How do I configure a custom schema name for Elementary so it lets me use a custom suffix
k
A message from kapa.ai
You can give Elementary its own schema (with whatever suffix you want) using dbt’s custom schema config plus, if needed, a custom 
generate_schema_name
macro. ### 1. Basic config in 
dbt_project.yml
If you just want a dedicated Elementary schema that follows dbt’s default pattern (
<target_schema>_elementary
), add:
Copy code
yml models:   elementary:     +schema: elementary
This makes Elementary models land in a schema like 
TARGETSCHEMA_elementary
. [[Elementary schema](https://docs.elementary-data.com/data-tests/dbt/dbt-package#page-title)] ### 2. Using a fully custom schema name If you want full control (e.g. 
elementary_data_observability
or a different suffix), override 
generate_schema_name
and use `meta.schema_name`:
Copy code
sql -- macros/generate_schema_name.sql {% macro generate_schema_name(custom_schema_name, node) -%}      {%- set default_schema = target.schema -%}     {% set config_meta = node.config.get('meta') %}      {% if config_meta and config_meta is mapping %}         {% set schema_name = config_meta.get('schema_name') %}         {% if schema_name and schema_name is string %}             {{ return(schema_name) }}         {% endif %}      {%- elif custom_schema_name is none -%}          {{ default_schema }}      {%- else -%}          {{ default_schema }}_{{ custom_schema_name | trim }}      {%- endif -%}  {%- endmacro %}
Then configure Elementary to use that explicit schema:
Copy code
yml models:   elementary:     +meta:       schema_name: "my_custom_elementary_suffix"
This will create all Elementary models in the schema 
my_custom_elementary_suffix
(no automatic prefix/suffix added). [[OSS FAQ – change schema](https://docs.elementary-data.com/oss/general/faq#page-title)] ### 3. If you already have a custom 
generate_schema_name
If your project already overrides 
generate_schema_name
, you can: - Either adapt it so that when 
node.package_name == 'elementary'
you return the exact schema you want, or - Use the 
meta.schema_name
pattern above and keep your existing logic for everything else. If you share your current 
generate_schema_name
macro and desired final schema (e.g. 
DBT_ELEMENTARY
, 
analytics_elementary
, etc.), I can show the minimal change needed.
v
@kapa.ai If a test fails on an upstream source, how do I see what gets affected downstream
k
A message from kapa.ai
You can use Elementary’s lineage and incident views to see exactly which downstream assets are affected. 1. Start from the failing test / incident - Open the incident for the failed upstream source or test in the Incidents page. - In the incident side panel you’ll see upstream and downstream assets for that test/model. [[Incidents](https://docs.elementary-data.com/cloud/features/alerts-and-incidents/incidents)] 2. Open lineage from the test/incident - From the test results / incident, click “view in lineage” to jump into the lineage graph focused on that asset. - In lineage, you can see all downstream tables and exposures (dashboards) that depend on the failing source. [[Triage incidents](https://docs.elementary-data.com/cloud/best-practices/triage-and-response#triage-incidents); [Lineage overview](https://docs.elementary-data.com/cloud/features/data-lineage/lineage)] 3. Use filters to focus on what matters - In the lineage view, use filters to highlight: - Only downstream assets of the failing source. - Assets by criticality, tags, owners, or type (e.g. exposures/BI). [[Lineage overview](https://docs.elementary-data.com/cloud/features/data-lineage/lineage); [Alert rules](https://docs.elementary-data.com/cloud/features/alerts-and-incidents/alert-rules)] - For column-level impact, expand the table and select a specific column to see only its downstream columns and assets. [[Column lineage](https://docs.elementary-data.com/cloud/features/data-lineage/column-level-lineage)] 4. For BI / dashboards - If you’ve connected a BI tool (Tableau, Looker, Power BI, etc.), Elementary extends lineage to the dashboard level, so you can see which dashboards/exposures are impacted by the failing upstream source. [[BI lineage](https://docs.elementary-data.com/cloud/integrations/bi/connect-bi-tool); [End-to-end lineage](https://docs.elementary-data.com/cloud/features/data-lineage/lineage)] This combination (incident → “view in lineage” → filters / column-level lineage / BI exposures) is the recommended way to see everything affected downstream by a failed upstream source.
v
@kapa.ai Why are my Elementary tables empty even though my dbt tests passed successfully?
k
A message from kapa.ai
There are several common reasons Elementary tables stay empty even when 
dbt test
succeeds. You can walk through these checks: --- ### 1. Elementary models not run (or run on a different target) Elementary tables are only created/populated after:
Copy code
bash dbt run --select elementary
If you ran this on a different 
--target
than your 
dbt test
, the hooks will try to insert into tables that don’t exist in the current target, so they stay empty. Run both on the same target, e.g.:
Copy code
bash dbt run --select elementary --target prod dbt test --target prod
[[Slack – non‑existent table](https://elementary-community.slack.com/archives/C02CTC89LAX/p1683764076675589?thread_ts=1683764076.675589&amp;cid=C02CTC89LAX); [Cloud troubleshoot](https://docs.elementary-data.com/cloud/guides/troubleshoot#page-title)] --- ### 2. No tests actually ran If 
dbt test
prints 
Nothing to do
, no tests executed, so 
elementary_test_results
will be empty. You must define dbt tests (generic or singular) on your models first. [[Empty tests](https://elementary-community.slack.com/archives/C02CTC89LAX/p1706612460738369?thread_ts=1706612460.738369&amp;cid=C02CTC89LAX)] --- ### 3. Wrong materialization (tables/views instead of incremental) 
elementary_test_results
must be incremental. If a global config like:
Copy code
yml models:   +materialized: table
overrides Elementary’s materializations, Elementary can’t append results and tables may stay empty or error. Fix by scoping that config to your own project only, not to all packages, then recreate Elementary models:
Copy code
bash dbt run --select elementary --full-refresh dbt test
[[OSS troubleshooting](https://docs.elementary-data.com/oss/general/troubleshooting#page-title); [Model_run_results issue](https://elementary-community.slack.com/archives/C02CTC89LAX/p1693825818714329?thread_ts=1693825818.714329&amp;cid=C02CTC89LAX)] --- ### 4. Hooks disabled via vars If any of these vars are set, Elementary won’t upload results, so tables stay empty even though tests pass:
Copy code
yml vars:   disable_run_results: true   disable_tests_results: true   disable_dbt_invocation_autoupload: true   disable_dbt_artifacts_autoupload: true
Remove or set them to 
false
and rerun:
Copy code
bash dbt run --select elementary dbt test
[[Disabled vars](https://elementary-community.slack.com/archives/C02CTC89LAX/p1699310045488249?thread_ts=1699310045.488249&amp;cid=C02CTC89LAX); [Hooks note](https://elementary-community.slack.com/archives/C02CTC89LAX/p1703824109693079?thread_ts=1703824109.693079&amp;cid=C02CTC89LAX)] --- ### 5. On‑run‑end hooks not firing (or failing earlier) Elementary uploads artifacts and test results in on‑run‑end hooks. If another on‑run‑end hook fails first, dbt won’t execute Elementary’s hook, so no rows are inserted. [[On‑run‑end behavior](https://elementary-community.slack.com/archives/C02CTC89LAX/p1722936228529189?thread_ts=1722936228.529189&amp;cid=C02CTC89LAX); [Hooks explanation](https://elementary-community.slack.com/archives/C02CTC89LAX/p1703824109693079?thread_ts=1703824109.693079&amp;cid=C02CTC89LAX)] Check your 
dbt.log
for lines like:
Copy code
text Elementary: Uploading test results. Elementary: Uploading run results.
If they’re missing, the hooks didn’t run successfully. --- ### 6. Schema / 
generate_schema_name
mismatch If you override 
generate_schema_name
, Elementary tables may be created in a different schema than you expect, and the hooks may be writing to (or looking for) another schema. This can make it look like tables are “empty” when you’re just querying the wrong schema. [[Schema override issue](https://elementary-community.slack.com/archives/C02CTC89LAX/p1712851926036669?thread_ts=1712851926.036669&amp;cid=C02CTC89LAX); [Another schema issue](https://elementary-community.slack.com/archives/C02CTC89LAX/p1685020975522279?thread_ts=1685020975.522279&amp;cid=C02CTC89LAX)] Confirm: - The schema configured for 
models: elementary:
in 
dbt_project.yml
. - Your custom 
generate_schema_name
macro doesn’t redirect Elementary models elsewhere. - You’re querying the same schema where 
dbt run --select elementary
actually built the models. --- ### 7. Incremental model got “stuck” Occasionally the incremental 
elementary_test_results
model can get into a bad state and stop appending. In that case, do a full refresh with the Elementary var:
Copy code
bash dbt run --select elementary --full-refresh --vars '{ "elementary_full_refresh": true }' dbt test
[[Incremental fix](https://elementary-community.slack.com/archives/C02CTC89LAX/p1736544876896189?thread_ts=1736544876.896189&amp;cid=C02CTC89LAX)] --- If you go through these and still see empty tables, the docs recommend checking 
elementary_test_results
existence/materialization and then sharing 
dbt.log
with the Elementary team. [[OSS troubleshooting](https://docs.elementary-data.com/oss/general/troubleshooting#page-title); [Cloud troubleshoot](https://docs.elementary-data.com/cloud/guides/troubleshoot#page-title)]
v
@kapa.ai How do I set up a dimension anomaly test to monitor the distribution of a 'category' column
k
A message from kapa.ai
To monitor the distribution of a 
category
column over time, use the 
elementary.dimension_anomalies
test with 
dimensions
set to that column. ### Basic setup (table-level test) In your 
schema.yml
(or equivalent):
Copy code
yml version: 2  models:   - name: your_model_name     config:       elementary:         timestamp_column: event_time   # recommended, but optional     data_tests:       - elementary.dimension_anomalies:           arguments:             dimensions:               - category             # optional filters & time config:             # where_expression: "category is not null"             # time_bucket:             #   period: day             #   count: 1
This test: - Counts rows grouped by 
category
. - Monitors how the frequency/distribution of each 
category
value changes over time. - Works best when 
category
has low cardinality. [[Dimension anomalies intro](https://docs.elementary-data.com/data-tests/anomaly-detection-tests/dimension-anomalies); [Dimensions config](https://docs.elementary-data.com/data-tests/anomaly-detection-configuration/dimensions)] ### Minimal version without timestamp If you don’t have a timestamp column:
Copy code
yml models:   - name: your_model_name     data_tests:       - elementary.dimension_anomalies:           arguments:             dimensions:               - category
Without 
timestamp_column
, Elementary monitors total row counts per 
category
(no time buckets). [[Add tests examples](https://docs.elementary-data.com/data-tests/add-elementary-tests#adding-tests-examples); [Dimension anomalies behavior](https://docs.elementary-data.com/data-tests/anomaly-detection-tests/dimension-anomalies)] Run it with:
Copy code
bash dbt test -m your_model_name
If you share your model name and timestamp column (if any), I can adapt this snippet exactly to your case.
v
@kapa.ai Does the open-source CLI version support column-level lineage, or is that a Cloud-only feature?
k
A message from kapa.ai
Column-level lineage is a Cloud-only feature today. - Elementary Cloud provides column-level lineage from source through dbt models up to BI tools/dashboards. [[Column lineage](https://docs.elementary-data.com/cloud/features/data-lineage/column-level-lineage); [Slack confirmation](https://elementary-community.slack.com/archives/C02CTC89LAX/p1763381285349839?thread_ts=1763381285.349839&amp;cid=C02CTC89LAX)] - The open-source (OSS) / CLI version only supports table-level lineage; column-level lineage is explicitly stated as available only in the Cloud Platform. [[OSS vs Cloud](https://docs.elementary-data.com/cloud/cloud-vs-oss#page-title); [Slack – Cloud only](https://elementary-community.slack.com/archives/C02CTC89LAX/p1714132214211679?thread_ts=1714132214.211679&amp;cid=C02CTC89LAX)]
v
@kapa.ai How do I adjust the sensitivity of a volume anomaly test to ignore fluctuations under 10%?
k
A message from kapa.ai
You can keep the anomaly test active but ignore changes smaller than 10% by using the 
ignore_small_changes
configuration on your 
elementary.volume_anomalies
test. 
ignore_small_changes
makes the test fail only if both: 1. The value is statistically anomalous (z‑score based), and 2. The change vs. the training-period mean is larger than your percent thresholds. [[ignore_small_changes](https://docs.elementary-data.com/data-tests/anomaly-detection-configuration/ignore_small_changes)] To ignore fluctuations under 10% (for both spikes and drops), configure:
Copy code
yml models:   - name: your_model     data_tests:       - elementary.volume_anomalies:           arguments:             ignore_small_changes:               spike_failure_percent_threshold: 10               drop_failure_percent_threshold: 10
With this setup, a volume anomaly will only fail if: - It’s flagged as anomalous by the model, and - It’s more than 10% higher (spike) or 10% lower (drop) than the mean in the training period. [[ignore_small_changes](https://docs.elementary-data.com/data-tests/anomaly-detection-configuration/ignore_small_changes); [volume_anomalies config](https://docs.elementary-data.com/data-tests/anomaly-detection-tests/volume-anomalies#test-configuration)]
v
@kapa.ai My on-run-end hook is failing with a permissions error in BigQuery what do I do
k
A message from kapa.ai
This usually means Elementary’s 
on-run-end
hook is trying to read/write tables or schemas that your BigQuery user doesn’t have access to. The fix is either to grant the right permissions or to disable Elementary’s artifact uploads where you don’t need them. ### 1. Check what the error is about From the examples in BigQuery/Snowflake, the failing objects are usually: - Elementary artifacts tables: 
DBT_MODELS
, 
DBT_TESTS
, 
DBT_RUN_RESULTS
, 
DATA_MONITORING_METRICS
, etc. [[Snowflake perms example](https://elementary-community.slack.com/archives/C02CTC89LAX/p1682612682384909?thread_ts=1682612682.384909&amp;cid=C02CTC89LAX); [Artifacts upload explanation](https://elementary-community.slack.com/archives/C02CTC89LAX/p1682612682384909?thread_ts=1682612682.384909&amp;cid=C02CTC89LAX)] - Or the Elementary schema itself (e.g. 
elementary
/ 
*_elementary
). [[Insufficient privileges elementary tables](https://elementary-community.slack.com/archives/C02CTC89LAX/p1695999713662539?thread_ts=1695999713.662539&amp;cid=C02CTC89LAX)] Look at the exact table/schema in the error message and confirm your dbt service account can: - Read all project schemas - Write to the Elementary schema - Create the Elementary schema (or you create it up front) [[Elementary in prod – permissions](https://docs.elementary-data.com/oss/deployment-and-configuration/elementary-in-production#page-title)] If this is a prod job, the recommended fix is to grant those permissions. --- ### 2. If you don’t want Elementary to run in this job (CI, seed, non‑prod) You can disable Elementary’s models or its artifact uploads so the 
on-run-end
hook doesn’t try to touch those tables. #### Option A – Disable Elementary models for certain targets In `dbt_project.yml`:
Copy code
yml models:   elementary:     +schema: "elementary"     +enabled: "{{ target.name in ['prod'] }}"
This runs Elementary only when 
target.name == 'prod'
. [[Enable only in prod](https://elementary-community.slack.com/archives/C02CTC89LAX/p1678899522071399?thread_ts=1678899522.071399&amp;cid=C02CTC89LAX)] For 
dbt seed
specifically, you can also disable Elementary when the command is `seed`:
Copy code
yml models:   elementary:     +enabled: "{{ flags.WHICH != 'seed' }}"     +schema: elementary
[[Disable for seed](https://elementary-community.slack.com/archives/C02CTC89LAX/p1682612682384909?thread_ts=1682612682.384909&amp;cid=C02CTC89LAX)] #### Option B – Keep models, but turn off artifact uploads in this env If you use Elementary only in prod and want CI/non‑prod to skip uploads, set vars:
Copy code
yml vars:   disable_dbt_artifacts_autoupload: true   disable_run_results: true   disable_tests_results: true   disable_dbt_invocation_autoupload: "{{ target.name != 'prod' }}"
This stops the 
on-run-end
hook from updating 
DATA_MONITORING_METRICS
, 
DBT_MODELS
, etc., which removes the permission requirement in those jobs. [[Vars to disable uploads](https://elementary-community.slack.com/archives/C02CTC89LAX/p1697606690660449?thread_ts=1697606690.660449&amp;cid=C02CTC89LAX)] --- ### 3. If the error is about accessing INFORMATION_SCHEMA / external projects If the error mentions 
INFORMATION_SCHEMA
or datasets in other projects you don’t control, you can temporarily disable the metadata models:
Copy code
yml models:   elementary:     edr:       dbt_artifacts:         dbt_columns:           +enabled: false       metadata_store:         +enabled: false
These are optional models; disabling them avoids information‑schema permissions issues. [[Disable metadata_store/dbt_columns](https://elementary-community.slack.com/archives/C02CTC89LAX/p1684375323318529?thread_ts=1684375323.318529&amp;cid=C02CTC89LAX)] --- If you paste the exact BigQuery error text (table/schema name and role), I can point to the minimal change: grant a specific permission vs. disable a specific Elementary piece in that environment.
v
@kapa.ai How do I route alerts for models with a specific tag to a specific Slack channel?
k
A message from kapa.ai
You have two main options in the OSS CLI, both supported by the docs and Slack threads: --- ## Option 1 – Use tags with 
edr monitor
selectors (per‑channel runs) Tag your models in dbt, then run 
edr monitor
once per tag/channel:
Copy code
yml models:   - name: my_finance_model     tags: ["finance"]
Then route alerts for that tag to a specific channel:
Copy code
bash edr monitor --select tag:finance --slack-channel-name finance_data_ops edr monitor --select tag:marketing --slack-channel-name marketing_data_ops
Each command sends alerts only for models/tests matching the tag selector to the given Slack channel. [[Tag routing example](https://elementary-community.slack.com/archives/C02CTC89LAX/p1690447538330909?thread_ts=1690447538.330909&amp;cid=C02CTC89LAX); [Selectors & channels](https://elementary-community.slack.com/archives/C02CTC89LAX/p1699185769367119?thread_ts=1699185769.367119&amp;cid=C02CTC89LAX)] > Note: run these 
edr monitor
commands sequentially, not in parallel, to avoid locks on the alerts status table. [[Selectors note](https://elementary-community.slack.com/archives/C02CTC89LAX/p1690447538330909?thread_ts=1690447538.330909&amp;cid=C02CTC89LAX)] --- ## Option 2 – Configure custom channels in dbt metadata (no per‑tag run needed) Instead of routing by tag at runtime, you can hard‑wire the Slack channel in dbt config using 
meta.channel
. For example, if your project folders align with domains:
Copy code
yml models:   finance:     +meta:       channel: finance_data_ops   marketing:     +meta:       channel: marketing_data_ops
All model and test alerts under 
models/finance
go to 
#finance_data_ops
, etc. [[Folder‑based channels](https://elementary-community.slack.com/archives/C02CTC89LAX/p1699185769367119?thread_ts=1699185769.367119&amp;cid=C02CTC89LAX); [Custom channel docs](https://docs.elementary-data.com/oss/guides/alerts/alerts-configuration#alerts-distribution)] You can also set it at the test path level:
Copy code
yml tests:   finance:     +meta:       channel: finance_data_ops
With this approach you run a single 
edr monitor
and Elementary routes alerts based on the configured 
meta.channel
. [[Routing explanation](https://elementary-community.slack.com/archives/C02CTC89LAX/p1699185769367119?thread_ts=1699185769.367119&amp;cid=C02CTC89LAX)] --- If you tell me your exact tag (e.g. 
tag:finance
) and desired Slack channel name, I can give you the precise 
edr monitor
command or 
dbt_project.yml
snippet.
v
@kapa.ai I added the dbt package to packages.yml, how do Imaterialize the Elementary tables in Snowflake?
k
A message from kapa.ai
To materialize the Elementary tables in Snowflake after adding the dbt package, you need to: 1. Configure the Elementary models’ schema in 
dbt_project.yml
 This gives Elementary its own schema (usually 
<target_schema>_elementary
):
Copy code
yml    models:      ## your project models      my_project:        # your configs here       ## Elementary models      elementary:        +schema: "elementary"
Depending on your 
generate_schema_name
macro, this will create the schema 
JAFFLE_SHOP_STAGE_ELEMENTARY
or similar in Snowflake. [[Elementary schema](https://docs.elementary-data.com/data-tests/dbt/dbt-package#page-title); [Quickstart dbt package](https://docs.elementary-data.com/data-tests/dbt/quickstart-package#page-title)] 2. Make sure global materializations don’t override Elementary If you have a global config like:
Copy code
yml    models:      my_project:        materialized: "{{ 'table' if target.name == 'prod-cloud' else 'view' }}"
keep it scoped to your project only, not to all models, so Elementary can keep its own (mostly incremental) materializations. [[Quickstart dbt package](https://docs.elementary-data.com/data-tests/dbt/quickstart-package#page-title)] 3. Run the Elementary models From your dbt project:
Copy code
bash    dbt deps    dbt run --select elementary
This will create/materialize the Elementary tables (e.g. 
ELEMENTARY_TEST_RESULTS
, 
DBT_RUN_RESULTS
, etc.) in your Snowflake warehouse under the configured schema. [[Quickstart dbt package](https://docs.elementary-data.com/data-tests/dbt/quickstart-package#page-title)] 4. (Optional) Full refresh of Elementary incremental models If you later change materialization settings or need to rebuild the tables:
Copy code
bash    dbt run --select elementary --full-refresh --vars '{"elementary_full_refresh": "true"}'
[[OSS FAQ – full refresh](https://docs.elementary-data.com/oss/general/faq#page-title)] After this, running 
dbt test
will start populating those Snowflake tables with test results and artifacts.
v
@kapa.ai What is the difference between freshness_anomalies and native dbt source_freshness
k
A message from kapa.ai
Elementary’s 
freshness_anomalies
and dbt’s native 
source freshness
solve related but different problems and work differently. ### 1. What they monitor - dbt 
source freshness
 - Checks if a source is fresh compared to a fixed threshold you define (e.g. “fail if data is older than 6 hours”). - Implemented only for sources, not models. [[Slack – dbt source freshness only for sources](https://elementary-community.slack.com/archives/C02CTC89LAX/p1696346623801349?thread_ts=1696346623.801349&amp;cid=C02CTC89LAX)] - Elementary 
freshness_anomalies
 - Monitors the time between updates over time and looks for anomalies in that pattern (statistical / z‑score based). [[Freshness anomalies intro](https://docs.elementary-data.com/data-tests/introduction#page-title); [freshness_anomalies behavior](https://docs.elementary-data.com/data-tests/anomaly-detection-tests/freshness-anomalies)] - Works on both sources and models. [[Slack – works on sources and models](https://elementary-community.slack.com/archives/C02CTC89LAX/p1696346623801349?thread_ts=1696346623.801349&amp;cid=C02CTC89LAX)] ### 2. Threshold vs anomaly‑based - dbt 
source freshness
 - You define explicit SLAs (e.g. 
warn_after
, 
error_after
); dbt simply checks if the latest timestamp exceeds those limits. - This is a threshold test. - 
freshness_anomalies
 - Does not use a user‑defined threshold like “older than 38 hours”. - Splits data into time buckets (daily by default), computes the max gap without updates per bucket, then compares recent buckets to historical ones; it fails only if the recent freshness is anomalous vs history. [[freshness_anomalies behavior](https://docs.elementary-data.com/data-tests/anomaly-detection-tests/freshness-anomalies)] - Elementary explicitly notes that if you want a single “older than X hours/days” rule, that’s not what 
freshness_anomalies
does; you’d use dbt source freshness or another threshold‑style test instead. [[Slack – threshold vs anomaly](https://elementary-community.slack.com/archives/C02CTC89LAX/p1683294589948059?thread_ts=1683294589.948059&amp;cid=C02CTC89LAX); [Slack – stale data & generic test](https://elementary-community.slack.com/archives/C02CTC89LAX/p1684401442974139?thread_ts=1684401442.974139&amp;cid=C02CTC89LAX)] ### 3. Behavior on rarely updated / already‑stale tables - If a table hasn’t been updated for a long time and that’s already its “normal” state: - 
freshness_anomalies
may pass, because all buckets look similarly stale, so there’s no anomaly. [[Slack – month‑old data still passing](https://elementary-community.slack.com/archives/C02CTC89LAX/p1684401442974139?thread_ts=1684401442.974139&amp;cid=C02CTC89LAX); [Slack – table not updated in a month](https://elementary-community.slack.com/archives/C02CTC89LAX/p1686242948441989?thread_ts=1686242948.441989&amp;cid=C02CTC89LAX)] - dbt 
source freshness
would still fail if the latest timestamp is older than your configured threshold. ### 4. How results appear in Elementary - 
freshness_anomalies
is a dbt test, so it runs with 
dbt test
and its results appear in the Elementary report/UI as an anomaly test. [[Add tests – freshness anomalies](https://docs.elementary-data.com/data-tests/add-elementary-tests#page-title)] - `dbt source freshness`: - Runs via 
dbt source freshness
. - Historically, it did not show up as a regular test in the Elementary report because it doesn’t return a standard dbt test result object; Elementary has used it mainly for alerts, with an open issue to surface it like other tests. [[Slack – not revealed in report](https://elementary-community.slack.com/archives/C02CTC89LAX/p1682416734937579?thread_ts=1682416734.937579&amp;cid=C02CTC89LAX); [Slack – implementation difference](https://elementary-community.slack.com/archives/C02CTC89LAX/p1682503992494309?thread_ts=1682503992.494309&amp;cid=C02CTC89LAX)] ### 5. When to use which - Use dbt 
source freshness
 when: - You have a clear SLA like “error if older than 6 hours / 1 day”. - You want deterministic pass/fail based on that SLA. - Use 
elementary.freshness_anomalies
 when: - You want Elementary to learn the normal update pattern and alert only when it deviates (e.g. daily table stops updating, or weekly table suddenly lags much longer than usual). - You want the same freshness logic on models as well as sources. [[Freshness anomalies description](https://docs.elementary-data.com/data-tests/introduction#page-title); [Slack – works on models and sources](https://elementary-community.slack.com/archives/C02CTC89LAX/p1696346623801349?thread_ts=1696346623.801349&amp;cid=C02CTC89LAX)] If you describe your update pattern and whether you prefer strict SLAs or anomaly‑based alerts, I can suggest a concrete config (or combination) that fits.
40 Views
Open in Slack
PreviousNext
Powered by