Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

feat: bind lambda with created sm endpoints #18

Merged
merged 1 commit into from
Aug 9, 2023
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 10 additions & 2 deletions src/api-stack.ts
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,9 @@ interface apiStackProps extends StackProps {
_vpc: ec2.Vpc;
_securityGroup: ec2.SecurityGroup;
_domainEndpoint: string;
_crossEndPoint: string;
_embeddingEndPoint: string;
_instructEndPoint: string;
}

export class LLMApiStack extends NestedStack {
Expand Down Expand Up @@ -42,7 +45,10 @@ export class LLMApiStack extends NestedStack {
securityGroups: [_securityGroup],
architecture: Architecture.X86_64,
environment: {
aos_endpoint: _domainEndpoint,
aos_endpoint: _domainEndpoint,
llm_endpoint: props._crossEndPoint,
embedding_endpoint: props._embeddingEndPoint,
cross_endpoint: props._crossEndPoint,
},
});

Expand Down Expand Up @@ -76,8 +82,10 @@ export class LLMApiStack extends NestedStack {
architecture: Architecture.X86_64,
environment: {
document_bucket: _S3Bucket.bucketName,
// embeddings_model_endpoint_name: TBD, wait for in sm-stack.ts
opensearch_cluster_domain: _domainEndpoint,
llm_endpoint: props._instructEndPoint,
embedding_endpoint: props._embeddingEndPoint,
cross_endpoint: props._crossEndPoint,
},
});

Expand Down
12 changes: 6 additions & 6 deletions src/llm-stack.ts
Original file line number Diff line number Diff line change
Expand Up @@ -15,9 +15,9 @@ interface llmStackProps extends StackProps {
}

export class LLMStack extends NestedStack {
_crossEndPointName;
_embeddingEndPointName;
_instructEndPointName;
_crossEndPoint;
_embeddingEndPoint;
_instructEndPoint;

constructor(scope: Construct, id: string, props: llmStackProps) {
super(scope, id, props);
Expand Down Expand Up @@ -99,7 +99,7 @@ export class LLMStack extends NestedStack {
endpointName: 'cross-endpoint',
});

this._crossEndPointName = crossEndpoint.endpointName;
this._crossEndPoint = crossEndpoint.endpointName;

// EMBEDDING MODEL
// Create model, BucketDeployment construct automatically handles dependencies to ensure model assets uploaded before creating the model in this.region
Expand Down Expand Up @@ -133,7 +133,7 @@ export class LLMStack extends NestedStack {
endpointName: 'embedding-endpoint',
});

this._embeddingEndPointName = embeddingEndpoint.endpointName;
this._embeddingEndPoint = embeddingEndpoint.endpointName;

// INSTRUCT MODEL
// Create model, BucketDeployment construct automatically handles dependencies to ensure model assets uploaded before creating the model in this.region
Expand Down Expand Up @@ -167,6 +167,6 @@ export class LLMStack extends NestedStack {
endpointName: 'instruct-endpoint',
});

this._instructEndPointName = instructEndpoint.endpointName;
this._instructEndPoint = instructEndpoint.endpointName;
}
}
38 changes: 23 additions & 15 deletions src/main.ts
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,17 @@ export class RootStack extends Stack {
default: 'llm-rag',
});

// This assest stack is to mitigate issue that the model assets in s3 bucket can't be located immediately to create sagemaker model
const _AssetsStack = new AssetsStack(this, 'assets-stack', {_s3ModelAssets:_S3ModelAssets.valueAsString, env:process.env});
const _LLMStack = new LLMStack(this, 'llm-stack', {
_s3ModelAssets:_S3ModelAssets.valueAsString,
_crossCodePrefix:_AssetsStack._crossCodePrefix,
_embeddingCodePrefix:_AssetsStack._embeddingCodePrefix,
_instructCodePrefix:_AssetsStack._instructCodePrefix,
env:process.env
});
_LLMStack.addDependency(_AssetsStack);

const _VpcStack = new VpcStack(this, 'vpc-stack', {env:process.env});

const _OsStack = new OpenSearchStack(this,'os-stack', {_vpc:_VpcStack._vpc, _securityGroup:_VpcStack._securityGroup});
Expand All @@ -32,32 +43,29 @@ export class RootStack extends Stack {
_Ec2Stack.addDependency(_VpcStack);
_Ec2Stack.addDependency(_OsStack);

const _ApiStack = new LLMApiStack(this, 'api-stack', {_vpc:_VpcStack._vpc, _securityGroup:_VpcStack._securityGroup, _domainEndpoint:_OsStack._domainEndpoint, env:process.env});
const _ApiStack = new LLMApiStack(this, 'api-stack', {
_vpc:_VpcStack._vpc,
_securityGroup:_VpcStack._securityGroup,
_domainEndpoint:_OsStack._domainEndpoint,
_crossEndPoint: _LLMStack._crossEndPoint ?? '',
_embeddingEndPoint:_LLMStack._embeddingEndPoint || '',
_instructEndPoint:_LLMStack._instructEndPoint || '',
env:process.env
});
_ApiStack.addDependency(_VpcStack);
_ApiStack.addDependency(_OsStack);

const _DynamoDBStack = new DynamoDBStack(this, 'ddb-stack', {_vpc:_VpcStack._vpc, _securityGroup:_VpcStack._securityGroup, _domainEndpoint:_OsStack._domainEndpoint, env:process.env});
_DynamoDBStack.addDependency(_VpcStack);
_DynamoDBStack.addDependency(_OsStack);

// This assest stack is to mitigate issue that the model assets in s3 bucket can't be located immediately to create sagemaker model
const _AssetsStack = new AssetsStack(this, 'assets-stack', {_s3ModelAssets:_S3ModelAssets.valueAsString, env:process.env});
const _LLMStack = new LLMStack(this, 'llm-stack', {
_s3ModelAssets:_S3ModelAssets.valueAsString,
_crossCodePrefix:_AssetsStack._crossCodePrefix,
_embeddingCodePrefix:_AssetsStack._embeddingCodePrefix,
_instructCodePrefix:_AssetsStack._instructCodePrefix,
env:process.env
});
_LLMStack.addDependency(_AssetsStack);

new CfnOutput(this, 'VPC',{value:_VpcStack._vpc.vpcId});
new CfnOutput(this, 'OpenSearch Endpoint',{value:_OsStack._domainEndpoint});
// contatenate the outputs from the ec2 stack with port 8081 and prefix _dashboards
new CfnOutput(this, 'OpenSearch Dashboard',{value:`${_Ec2Stack._publicIP}:8081/_dashboards`});
new CfnOutput(this, 'Cross Model Endpoint',{value:_LLMStack._crossEndPointName || 'No Cross Endpoint Created'});
new CfnOutput(this, 'Embedding Model Endpoint',{value:_LLMStack._embeddingEndPointName || 'No Embedding Endpoint Created'});
new CfnOutput(this, 'Instruct Model Endpoint',{value:_LLMStack._instructEndPointName || 'No Instruct Endpoint Created'});
new CfnOutput(this, 'Cross Model Endpoint',{value:_LLMStack._crossEndPoint || 'No Cross Endpoint Created'});
new CfnOutput(this, 'Embedding Model Endpoint',{value:_LLMStack._embeddingEndPoint || 'No Embedding Endpoint Created'});
new CfnOutput(this, 'Instruct Model Endpoint',{value:_LLMStack._instructEndPoint || 'No Instruct Endpoint Created'});
}
}

Expand Down