Skip to content

Commit

Permalink
docs(ingest): clarify bigquery-beta multiproject setup (datahub-proje…
Browse files Browse the repository at this point in the history 
…ct#6071)
  • Loading branch information
@hsheth2 @shirshanka
hsheth2 authored and shirshanka committed Sep 29, 2022
1 parent 4c999f4 commit b804516
Show file tree
Hide file tree
Showing 2 changed files with 63 additions and 56 deletions.
112 changes: 56 additions & 56 deletions metadata-ingestion/docs/sources/bigquery/bigquery-beta_pre.md
View file
Original file line number Diff line number Diff line change
@@ -1,51 +1,62 @@
### Prerequisites

#### Create a datahub profile in GCP
1. Create a custom role for datahub as per [BigQuery docs](https://cloud.google.com/iam/docs/creating-custom-roles#creating_a_custom_role)

1. Create a custom role for datahub as per [BigQuery docs](https://cloud.google.com/iam/docs/creating-custom-roles#creating_a_custom_role).
2. Grant the following permissions to this role:

:::info

If you have multiple projects in your BigQuery setup, the role should be granted these permissions in each of the projects.

:::

##### Basic Requirements (needs for metadata ingestion)
| permission | Description |
|------------|----------------------------------------------|
| `bigquery.datasets.get` | This needs to list datasets |
| `bigquery.datasets.getIamPolicy` | This needs to list datasets |
| `bigquery.jobs.create` | Needs to submit queries. |
| `bigquery.jobs.list` | Needs to check submitted queries status. |
| `bigquery.tables.get` | Needs to get metadata about Bigquery Tables. |
| `bigquery.tables.list` | Needs to list metadata about Bigquery Tables. |
| `bigquery.readsessions.create` | Needs to get resultset of queries. |
| `bigquery.readsessions.getData` | Needs to get resultset of queries. |
| `resourcemanager.projects.get` | Needs to get resultset of queries. |

| permission | Description |
| -------------------------------- | ----------------------------------------------------- |
| `bigquery.datasets.get` | Retrieve metadata about a dataset. |
| `bigquery.datasets.getIamPolicy` | Read a dataset's IAM permissions. |
| `bigquery.jobs.create` | Run jobs (e.g. queries) within the project. |
| `bigquery.jobs.list` | Manage the queries that the service account has sent. |
| `bigquery.tables.list` | List BigQuery tables. |
| `bigquery.tables.get` | Retrieve metadata for a table. |
| `bigquery.readsessions.create` | Create a session for streaming large results. |
| `bigquery.readsessions.getData` | Get data from the read session. |
| `resourcemanager.projects.get` | Retrieve project names and metadata. |

##### Lineage/usage generation requirements

Additional requirements needed on the top of the basic requirements.
If you want to get lineage from multiple projects you have to grant this permission
for each of them.

| permission | Description |
|------------|-------------------------------------------------------------------------------------|
| `bigquery.jobs.listAll` | Needs for lineage generation and usage to see all the queries were run on a project |
| `logging.logEntries.list` | Needs for lineage generation via GCP logging |
| `logging.privateLogEntries.list` | Needs for lineage generation via GCP logging |
| permission | Description |
| -------------------------------- | ------------------------------------------------------------------------------------------------------------ |
| `bigquery.jobs.listAll` | List all jobs (queries) submitted by any user. |
| `logging.logEntries.list` | Fetch log entries for lineage/usage data. Not required if `use_exported_bigquery_audit_metadata` is enabled. |
| `logging.privateLogEntries.list` | Fetch log entries for lineage/usage data. Not required if `use_exported_bigquery_audit_metadata` is enabled. |

##### Profiling requirements

Additional requirements needed on the top of the basic requirements.

| permission | Description |
|------------|---------------------------------------------------------------------------------------------------|
| `bigquery.tables.getData` | profiler needs to access data to do the profiling |
| `bigquery.tables.create` | It needs to create temporary tables to profile partitioned/sharded tables. See below for details. |
| `bigquery.tables.delete` | It needs to create temporary tables to profile partitioned/sharded tables. See below for details. |
| permission | Description |
| ------------------------- | ----------------------------------------------------------------------------------------- |
| `bigquery.tables.getData` | Access table data to do the profiling. |
| `bigquery.tables.create` | Create temporary tables when profiling partitioned/sharded tables. See below for details. |
| `bigquery.tables.delete` | Delete temporary tables when profiling partitioned/sharded tables. See below for details. |

Profiler creates temporary tables to profile partitioned/sharded tables and that is why it needs table create/delete privilege.
Use `profiling.bigquery_temp_table_schema` to restrict to one specific dataset the create/delete permission

#### Create a service account

1. Setup a ServiceAccount as per [BigQuery docs](https://cloud.google.com/iam/docs/creating-managing-service-accounts#iam-service-accounts-create-console)
and assign the previously created role to this service account.
and assign the previously created role to this service account.
2. Download a service account JSON keyfile.
Example credential file:

```json
{
"type": "service_account",
Expand All @@ -60,22 +71,27 @@ and assign the previously created role to this service account.
"client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/test%suppproject-id-1234567.iam.gserviceaccount.com"
}
```

3. To provide credentials to the source, you can either:

Set an environment variable:
$ export GOOGLE_APPLICATION_CREDENTIALS="/path/to/keyfile.json"

*or*
```sh
$ export GOOGLE_APPLICATION_CREDENTIALS="/path/to/keyfile.json"
```

Set credential config in your source based on the credential json file. For example:
_or_

```yml
credential:
project_id: project-id-1234567
private_key_id: "d0121d0000882411234e11166c6aaa23ed5d74e0"
private_key: "-----BEGIN PRIVATE KEY-----\nMIIyourkey\n-----END PRIVATE KEY-----\n"
client_email: "[email protected]"
client_id: "123456678890"
```
Set credential config in your source based on the credential json file. For example:

```yml
credential:
project_id: project-id-1234567
private_key_id: "d0121d0000882411234e11166c6aaa23ed5d74e0"
private_key: "-----BEGIN PRIVATE KEY-----\nMIIyourkey\n-----END PRIVATE KEY-----\n"
client_email: "[email protected]"
client_id: "123456678890"
```
### Lineage Computation Details
Expand All @@ -91,13 +107,13 @@ Profiling can profile normal/partitioned and sharded tables as well but due to p

If limit/offset parameter is set or partitioning partitioned or sharded table Great Expectation (the profiling framework we use) needs to create temporary
views. By default, these views are created in the schema where the profiled table is but you can control to create all these
tables into a predefined schema by setting `profiling.bigquery_temp_table_schema` property.
tables into a predefined schema by setting `profiling.bigquery_temp_table_schema` property.
Temporary tables are removed after profiling.

```yaml
profiling:
enabled: true
bigquery_temp_table_schema: my-project-id.my-schema-where-views-can-be-created
profiling:
enabled: true
bigquery_temp_table_schema: my-project-id.my-schema-where-views-can-be-created
```

:::note
Expand All @@ -106,22 +122,6 @@ Due to performance reasons, we only profile the latest partition for Partitioned
You can set partition explicitly with `partition.partition_datetime` property if you want. (partition will be applied to all partitioned tables)
:::

### Working with multi-project GCP setups

Sometimes you may have multiple GCP project with one only giving you view access rights and other project where you have view/modify rights.

The GCP roles with which this setup has been tested are as follows
- Storage Project
- BigQuery Data Viewer
- BigQuery Metadata Viewer
- Logs Viewer
- Private Logs Viewer
- Compute Project
- BigQuery Admin
- BigQuery Data Editor
- BigQuery Job User

If you are using `use_exported_bigquery_audit_metadata = True` then make sure you prefix the datasets in `bigquery_audit_metadata_datasets` with storage project id.

### Caveats
- For Materialized views lineage is dependent on logs being retained. If your GCP logging is retained for 30 days (default) and 30 days have passed since the creation of the materialized view we won't be able to get lineage for them.

- For materialized views, lineage is dependent on logs being retained. If your GCP logging is retained for 30 days (default) and 30 days have passed since the creation of the materialized view we won't be able to get lineage for them.
7 changes: 7 additions & 0 deletions metadata-ingestion/src/datahub/ingestion/source/bigquery_v2/bigquery_config.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -57,6 +57,13 @@ class BigQueryV2Config(BigQueryConfig):
description="Number of table queried in batch when getting metadata. This is a low leve config propert which should be touched with care. This restriction needed because we query partitions system view which throws error if we try to touch too many tables.",
)

# The inheritance hierarchy is wonky here, but these options need modifications.
project_id: Optional[str] = Field(
default=None,
description="[deprecated] Use project_id_pattern instead.",
)
storage_project_id: None = Field(default=None, exclude=True)

@root_validator(pre=False)
def profile_default_settings(cls, values: Dict) -> Dict:
# Extra default SQLAlchemy option for better connection pooling and threading.
Expand Down

0 comments on commit b804516

Please sign in to comment.