Compare commits

...
Author SHA1 Message Date
Rafael c8a4c1aae6 UPDATE: migrating maestro to backend nodepool 2025-02-20 13:46:30 -03:00
Rafael f48a35c7ce UPDATE: improve logging for table metadata 2025-02-20 08:40:48 -03:00
Rafael 831a03402e UPDATE: updating protospack for add table_schema for /catalog/register-dataset rout 2025-02-19 12:19:59 -03:00
Rafael 856f913ce6 Merge branch 'feat/catalog-external-datasets-using-api-call' of github.com:dadosfera/maestro into feat/catalog-external-datasets-using-api-call 2025-02-14 15:07:13 -03:00
Rafael dc617b6925 REFACTOR: Improve logging with additional metadata
- Enhanced logs to include more detailed request body information.
- Standardized log calls to include extra metadata, such as service name, environment, customer name, user ID, HTTP method, and request path.
- Improved error logging to provide better debugging insights.

Example log format:
EXTRA[{"service":{"name":"maestro","environment":"stg"},"customer_name":"dadosfera","user_id":"047824d8-0021-709a-bf15-39b8dd069aaf","method":"POST","path":"/catalog/register-dataset"}]

Related tickets or issues: [TECN-9441]
2025-02-14 15:04:22 -03:00
Rafael Santana 3f80062672 Merge branch 'beta' into feat/catalog-external-datasets-using-api-call 2025-02-14 10:14:01 -03:00
Rafael 07eb423814 UPDATE: update version of protospack 2025-02-14 10:12:20 -03:00
Rafael 721232448e FEAT: Enable dataset registration via API for external sources
As part of our hybrid approach, we need to allow dataset cataloging from outside our cloud environment.
Instead of relying on our GRPC APIs, we leveraged an existing REST API to handle this process.

- Updated Maestro to call Nimbus directly for dataset registration.
- Modified protospack to generate the necessary TypeScript interfaces.
- Added the `/catalog/register-dataset` endpoint to the API.

This change enables seamless dataset registration from external sources while maintaining compatibility with our existing infrastructure.

Related tickets or issues: [TECN-9441]
2025-02-14 10:08:45 -03:00
Marcos Rodrigues Silva 0b0d261aae Merge pull request #256 from dadosfera/fix/white-label
UPDATE: remove user agent
2025-02-13 13:56:37 -03:00
marcos-silva-rodrigues 87a8d741d7 UPDATE: remove user agent 2025-02-13 13:55:37 -03:00
10 changed files with 1873 additions and 1408 deletions
+35
View File
@@ -4787,6 +4787,41 @@
]
}
},
"/catalog/register-dataset": {
"post": {
"operationId": "CatalogController_registerDatasetWithMetadataRequest",
"parameters": [
{
"name": "dadosfera-lang",
"in": "header",
"required": false,
"schema": {
"enum": [
"pt-br",
"en-us"
],
"type": "string"
}
}
],
"responses": {
"201": {
"description": ""
}
},
"tags": [
"Catalog"
],
"security": [
{
"access-token": []
},
{
"access-token": []
}
]
}
},
"/terms-of-use/token": {
"get": {
"operationId": "TermsOfUseController_getToken",
+3 -1
View File
@@ -17,4 +17,6 @@ charts:
- name: maestro.open_customer_id
value: f239718a-a271-4ef9-ae7e-02a2f0f3aa6e
- name: maestro.open_group_id
value: 401573bb-334f-44b2-b30e-88d4cea31ae9
value: 401573bb-334f-44b2-b30e-88d4cea31ae9
- name: replicaCount
value: 2
+2 -1
View File
@@ -35,7 +35,8 @@ spec:
- key: application
operator: In
values:
- backend
- general
tolerations:
- key: "kubernetes.azure.com/scalesetpriority"
operator: "Equal"
+1 -1
View File
@@ -3,7 +3,7 @@
# Declare variables to be passed into your templates.
replicaCount: 3
hostname: maestro-temp.dadosfera.ai
hostname: maestro.dadosfera.ai
image:
repository: 611330257153.dkr.ecr.us-east-1.amazonaws.com/microservices/maestro_prd
pullPolicy: IfNotPresent
+1617 -1401
View File
File diff suppressed because it is too large Load Diff
+1 -1
View File
@@ -30,7 +30,7 @@
"@aws-sdk/client-secrets-manager": "^3.414.0",
"@dadosfera/dadosfera-logs": "^1.0.0-beta.4",
"@dadosfera/protospack": "2.5.3",
"@dadosfera/protospack-v2": "3.37.0-beta.5",
"@dadosfera/protospack-v2": "3.37.0-beta.7",
"@grpc/grpc-js": "^1.9.3",
"@grpc/proto-loader": "^0.7.9",
"@nestjs/cli": "^9.5.0",
+3
View File
@@ -5,6 +5,7 @@ import {
} from '@nestjs/microservices';
import { credentials } from '@grpc/grpc-js';
import { Catalog } from '@dadosfera/protospack-v2';
import { PlatformInterfaces } from '@dadosfera/protospack-v2';
const isLocalConnection =
process.env.PIFACTORY_URL.startsWith('pi-factory:') ||
@@ -19,11 +20,13 @@ export class CatalogClientConfiguration {
package: [
Catalog.ProtoPackages.ReadPackage,
Catalog.ProtoPackages.WritePackage,
PlatformInterfaces.ProtoPackages.WritePackage
],
credentials: isLocalConnection ? undefined : credentials.createSsl(),
protoPath: [
Catalog.ProtoPaths.ReadFilePath,
Catalog.ProtoPaths.WriteFilePath,
PlatformInterfaces.ProtoPaths.WriteFilePath
],
loader: {
keepCase: true,
+94
View File
@@ -13,6 +13,8 @@ import {
Put,
Query,
UseFilters,
HttpException,
HttpStatus,
} from '@nestjs/common';
import {
ApiCreatedResponse,
@@ -30,6 +32,7 @@ import { CatalogService } from './catalog.service';
import { DadosferaLogger } from '@dadosfera/dadosfera-logs';
import { PackTheMetadata } from 'src/utils/ PackTheMetadata';
import { RequestUser, User } from 'src/decorators/user.decorator';
import { lastValueFrom } from 'rxjs';
import {
BatchRemoveRlsRulesRequest,
GetDatasetCatalogTaskRes,
@@ -54,6 +57,7 @@ import {
AddRlsRuleRequest,
GetNimbusDashboardsRequest,
GetRlsRulesRequest,
RegisterDatasetWithMetatadaRequest,
} from '@dadosfera/protospack-v2/dist/lib/Catalog/interfaces/messages';
@ApiTags('Catalog')
@@ -743,4 +747,94 @@ export class CatalogController {
return JSON.parse(dashboards);
}
@Post('register-dataset')
@RequireSomePermission(
PERMISSIONS_GROUPS.CATALOG.permissions.CREATE,
PERMISSIONS_GROUPS.CATALOG.permissions.DATA_MANAGER
)
async registerDatasetWithMetadataRequest(
@Body() body: RegisterDatasetWithMetatadaRequest,
@User() user: RequestUser,
) {
const { customer_id, customer_name, user_id, username } = user;
const logMetadata = {
customer_name: customer_name,
user_id: user_id,
method: 'POST',
path: '/catalog/register-dataset',
};
try {
const metadata = PackTheMetadata({
customer_id,
customer_name,
user_id,
username,
});
this.logger.log(`Request from user ${user_id} for customer ${customer_name}`, logMetadata);
// Create table metadata
const tableMetadataBody = {
table_metadata: body.table_metadata,
info: {
customer: customer_name,
},
logMetadata: logMetadata
};
const table_metadata_id = await this.catalogService.createTableMetadata(
tableMetadataBody,
);
this.logger.info(`table_metadata_id: ${table_metadata_id}`, logMetadata);
// Create column metadata
const columnMetadataBody = {
column_metadata: body.column_metadata,
info: {
customer: customer_name,
},
};
this.logger.info(`Creating column metadata for table ${table_metadata_id}`, logMetadata);
const column_metadata_ids = await this.catalogService.createColumnMetadata(columnMetadataBody)
// Create data preview
const dataPreviewBody = {
data_preview: body.data_preview,
info: {
customer: customer_name,
},
};
this.logger.debug(`Creating data preview for table ${table_metadata_id}`, logMetadata);
const data_preview_id = await this.catalogService.createDataPreview(
dataPreviewBody,
);
// Catalog dataset item
this.logger.info(`Cataloging dataset item for table ${table_metadata_id}`, logMetadata);
await this.catalogService.catalogDatasetItem(table_metadata_id, metadata);
this.logger.info(`Dataset registration completed successfully for table ${table_metadata_id}`, logMetadata);
return {
message: 'Dataset registered successfully',
table_metadata_id: table_metadata_id,
column_metadata_ids: column_metadata_ids,
data_preview_id: data_preview_id,
};
} catch (error) {
this.logger.error(`Failed to register dataset. The following error occurred: ${error.response.data}`, logMetadata);
throw new HttpException(
{
message: 'Ocorreu um erro ao registrar o dataset',
error: error.message,
code: 'REGISTRATION_FAILED',
details: error.message,
},
HttpStatus.INTERNAL_SERVER_ERROR,
);
}
}
}
+117
View File
@@ -5,6 +5,11 @@ import {
ProtoServices,
Messages,
} from '@dadosfera/protospack-v2/dist/lib/Catalog';
import {
Messages as PlatformInterfaceMessages,
WriteService as PlatformInterfaceWriteService,
ProtoServices as PlatformInterfacesProtoServices,
} from '@dadosfera/protospack-v2/dist/lib/PlatformInterfaces';
import {
BadRequestException,
HttpException,
@@ -33,6 +38,7 @@ import {
class CatalogService implements OnModuleInit {
catalogReadService: ReadService.CatalogReadServices;
catalogWriteService: WriteService.CatalogWriteServices;
platformWriteService: PlatformInterfaceWriteService.PlatformInterfacesWriteServices;
logger: any;
constructor(
@Inject(DadosferaLogger)
@@ -54,9 +60,14 @@ class CatalogService implements OnModuleInit {
this.grpcClient.getService<WriteService.CatalogWriteServices>(
ProtoServices.CatalogWriteServices,
);
this.platformWriteService =
this.grpcClient.getService<PlatformInterfaceWriteService.PlatformInterfacesWriteServices>(
PlatformInterfacesProtoServices.PlatformInterfacesWriteServices,
);
}
_getNimbusUrl(body) {
this.logger.debug(`Body: ${JSON.stringify(body)}`);
const customer = body.info.customer.toLowerCase();
if (process.env.ENV === 'prd') {
@@ -420,6 +431,112 @@ class CatalogService implements OnModuleInit {
);
return res.dashboards;
}
async createTableMetadata(body: any): Promise<number> {
const nimbusUrl = this._getNimbusUrl(body);
this.logger.info(`Nimbus URL: ${nimbusUrl}`, {...body.logMetadata});
const endpoint = `${nimbusUrl}/api/catalog/table-metadata/`;
this.logger.info(`Creating table metadata for table ${body.table_metadata.table_name}`, {...body.logMetadata});
this.logger.info(`Using endpoint: ${endpoint}`, {...body.logMetadata});
this.logger.debug(`Payload: ${JSON.stringify(body.table_metadata)}`, {...body.logMetadata});
try {
const { data, status } = await axios.post(endpoint, {...body.table_metadata});
this.logger.info(
`Table metadata created successfully with status ${status} for table ${body.table_metadata.table_name}`,
{...body.logMetadata},
);
return data.id;
} catch (error) {
this.logger.error(
`Failed to create table metadata for table ${body.table_metadata.table_name} failed with status ${
error.response?.status
} because of ${JSON.stringify(error.response?.data) || error.message}`, {...body.logMetadata});
throw new Error(error.response?.data?.message || error.message);
}
}
async createColumnMetadata(body: any): Promise<number[]> {
const nimbusUrl = this._getNimbusUrl(body);
this.logger.info(`Nimbus URL: ${nimbusUrl}`, body.logMetadata);
const endpoint = `${nimbusUrl}/api/catalog/column-metadata/`;
try {
this.logger.info(`Creating column metadata for table ${body.column_metadata.table_name}`, {...body.logMetadata});
this.logger.info(`Using endpoint: ${endpoint}`, {...body.logMetadata});
this.logger.debug(`Payload: ${JSON.stringify(body.column_metadata)}`, {...body.logMetadata});
const { data, status } = await axios.post(endpoint, body.column_metadata);
this.logger.info(
`Column metadata created successfully with status ${status} for table ${body.column_metadata.table_name}`,
{...body.logMetadata},
);
return data.map((column) => column.id);
} catch (error) {
this.logger.error(
`Failed to create column metadata failed with status for table ${body.column_metadata.table_name} ${
error.response?.status
} because of ${error.response?.data || error.message}`, {...body.logMetadata});
throw new Error(error.response?.data?.message || error.message);
}
}
async createDataPreview(body: any): Promise<number> {
const nimbusUrl = this._getNimbusUrl(body);
this.logger.info(`Nimbus URL: ${nimbusUrl}`, {...body.logMetadata});
const endpoint = `${nimbusUrl}/api/catalog/data-preview/`;
this.logger.info(`Creating data preview for table ${body.data_preview.table_name}`, {...body.logMetadata});
this.logger.info(`Using endpoint: ${endpoint}`, {...body.logMetadata});
this.logger.debug(`Payload: ${JSON.stringify(body.data_preview)}`, {...body.logMetadata});
try {
const { data, status } = await axios.post(endpoint, body.data_preview);
this.logger.info(
`Data preview created successfully with status ${status} for table ${body.data_preview.table_name}`,
{...body.logMetadata},
);
return data.id;
} catch (error) {
this.logger.error(
`Failed to create data preview for table ${body.data_preview.table_name} failed with status ${
error.response?.status
} because of ${error.response?.data || error.message}`,
{...body.logMetadata},
);
throw new Error(error.response?.data?.message || error.message);
}
}
async catalogDatasetItem(table_metadata_id: number, metadata: Metadata) {
const customer_name_raw = metadata.get('customer_name');
const customer_name = customer_name_raw?.[0]?.toString();
if (!customer_name) {
throw new BadRequestException('Customer name not found in metadata');
}
const res = await lastValueFrom(
this.platformWriteService.CatalogDataAssets(
{
data_assets: [
{
data_asset_id: table_metadata_id.toString(),
customer_name: customer_name,
data_asset_type: 'dataset',
},
],
},
metadata,
),
);
return res;
}
}
export { CatalogService };
@@ -41,9 +41,6 @@ export class OpenDataController {
@Req()
request: Request,
) {
if (!request.headers['user-agent'].includes('WordPress/6.7.1; https://devsbm.dadosfera.io')) {
throw new ForbiddenException();
}
this.logger.info('createUser for open data');
this.logger.info('user-agent', request.headers['user-agent']);