I’ve got an issue with some dbt models not combining with their respective Snowflake tables. The issue is only affecting some tables (eg works for RAW_ANSWERS but not DIM_ANSWERS in the screenshot). It is also only affecting our production instance of Datahub, it works fine on dev instance ingesting the same data and using the same server version (0.10.2).
I can’t see why certain tables wouldn’t combine and don’t have many ideas for debugging. Does anyone have ideas for how to correct? I’ve tried rerunning the ingestions with dbt first and with Snowflake first but saw no change. I see no difference in naming or URNs between our prod and dev instances.
Are you able to upgrade to the newest version, 0.10.5? We had a bug with the creation of siblings (combining entities) that was fixed somewhat recently
urn:li:dataset:(urn:li:dataPlatform:snowflake,mydb.myschema.dim_answers,PROD)```
This is especially confusing as the urns and, from what I can tell, all the other configs, seem to the same on our dev instance of Datahub, but the dim_answers table and model do combine there.
Unfortunately I can’t easily upgrade the server version right now. The recipes are below, but I think the only thing changing between the runs is the rest endpoint.
dbt cloud
```source:
type: "dbt-cloud"
config:
token: x
account_id: x
project_id: x
job_id: x
metadata_endpoint: <https://metadata.emea.dbt.com/graphql>
target_platform: snowflake
stateful_ingestion:
enabled: false
env: PROD```
snowflake
```pipeline_name: snowflake__policy_domain
source:
type: snowflake
config:
account_id: x
warehouse: x
username: x
password: x
role: x
database_pattern:
allow:
- ^...$
- ...
schema_pattern:
allow:
- ...
profiling:
enabled: false
turn_off_expensive_profiling_metrics: true
ignore_start_time_lineage: true
stateful_ingestion:
enabled: false
remove_stale_metadata: false
env: PROD
transformers:
- type: pattern_add_dataset_terms
config:
semantics: PATCH
term_pattern:
rules:
...```