aws-serverless-samfarm
aws-serverless-samfarm copied to clipboard
Multiple Lambda functions from the sam.yml file possible?
I have a situation where I have multiple lambda functions within our API Gateway's API. I'm new to AWS and learning as much as I can. I am wondering, how would I change the template yaml (sam.yaml) so that we can deploy to any number of lambda functions and not just one as is the case in this lab. Would you create multiple index.js file per lambda function?
- so index1.js -> will map to first lambda function
- index2.js -> second lambda function;
- etc....
And how do I update the sam.yaml file to reflect this change?
Here's a snippet code of what I'm trying to accomplish:
Resources:
HelloWorld:
Type: AWS::Serverless::Function
Properties:
Handler: index.handler
Runtime: nodejs4.3
Role:
Fn::ImportValue:
!Join ['-', [!Ref 'ProjectId', !Ref 'AWS::Region', 'LambdaTrustRole']]
Events:
GetEvent:
Type: Api
Properties:
Path: /
Method: get
PostEvent:
Type: Api
Properties:
Path: /
Method: post
.... and more endpoints here to hit more lambda functions? So do I create multiple index.js to map to each individual lambda function?
Guys,
I have the same question.
Really trying to learn serverless and questions like these - on best practices - is making it super difficult. Would love some direction here.
any help
As far as I am aware, you can have multiple JS files, each containing a handler to a different Lambda. The files don't have to be in the same directory! You can also define the 2nd Lambda function in the same SAM template yaml file where you defined the 1st Lambda function.
AWSTemplateFormatVersion: '2010-09-09'
Transform: AWS::Serverless-2016-10-31
Description: Outputs the time
Resources:
TimeFunction:
Type: AWS::Serverless::Function
Properties:
Handler: firstsample/firstsample.handler # firstsample.js file is in firstsample direcotory
Role: !GetAtt BasicAWSLambdaRole.Arn
Runtime: nodejs6.10
CodeUri: ./ # Look for code in the same directory as the output SAM file
Events:
MyTimeApi:
Type: Api
Properties:
Path: /TimeResource
Method: GET
SecondSampleFunction:
Type: AWS::Serverless::Function
Properties:
Handler: secondsample.handler # didn't have to include secondsample directory
Role: !GetAtt BasicAWSLambdaRole.Arn
Runtime: nodejs6.10
CodeUri: ./secondsample # code is in the secondsample directory, located in same directory
Events:
MyTimeApi:
Type: Api
Properties:
Path: /TextResource
Method: GET
This is the top google result for "sam template multiple functions". Be nice to have airoVera's solution in the official docs if thats the way we should be doing this.
And how does this work with aws cloudformation package
and aws cloudformation deploy
? Do I need to run that for each lambda/CodeUri?
@red8888 what I do is having my repo structure as below. I used Jenkins pipeline to do 'npm install' for my lambda functions, FYR.
├── 00_DEVOPS-test1 │ ├── index.js │ ├── package.json │ ├── package-lock.json │ └── README.md ├── 00_DEVOPS-test2 │ ├── index.js │ ├── package.json │ ├── package-lock.json │ └── README.md ├── Jenkinsfile ├── packaged.yaml ├── README.md └── template.yaml
@red8888, using the @jairoVera approach, as both AWS::Serverless::Function
are on the same template only one sam package | aws cloudformation package
must be executed, same for the deploy
.
Regarding the first code if you have a main.js
that check the origin of the request and it can route it Post to index-1.js
and the Get to index-2.js
using the main.js
handler and directing to 2 "subhandlers"
@jairoVera do you know if that approach will generate 1 or 2 API Gateways?
It will generate two separate API gateways but the issue is (at least when generating a testing environment locally) if you create separate functions that have separate handler files both of the handler files will be written into the build folder for BOTH Lambda functions. I'm not sure if this would happen when deploying though.
It seems SAM is intended for one lambda per application. For multi-function applications, there is Nested application method.
SAM does support multiple functions, the right way to do it is with the sam build
command which pulls deps for all functions and creates a package that will be deployed as a "serverless application"
Maybe someone from the SAM team can reiterate this?
@deleugpn it will generate two API Gateways, but you can add the API Gateway resource and It will create just one
Transform: AWS::Serverless-2016-10-31
Description: Outputs the time
Resources:
ApiResource:
Type: AWS::Serverless::Api
Properties:
StageName: prod
TimeFunction:
Type: AWS::Serverless::Function
Properties:
Handler: firstsample/firstsample.handler # firstsample.js file is in firstsample direcotory
Role: !GetAtt BasicAWSLambdaRole.Arn
Runtime: nodejs6.10
CodeUri: ./ # Look for code in the same directory as the output SAM file
Events:
MyTimeApi:
Type: Api
Properties:
Path: /TimeResource
Method: GET
RestApiId: !Ref ApiResource
SecondSampleFunction:
Type: AWS::Serverless::Function
Properties:
Handler: secondsample.handler # didn't have to include secondsample directory
Role: !GetAtt BasicAWSLambdaRole.Arn
Runtime: nodejs6.10
CodeUri: ./secondsample # code is in the secondsample directory, located in same directory
Events:
MyTimeApi:
Type: Api
Properties:
Path: /TextResource
Method: GET
RestApiId: !Ref ApiResource```
And how about shared libraries and code?
You can use Layers
@jairoVera is life saver, I was banging on this problem for a week.
I have 13 serverless functions having go1.x and python run times and dependent to each others, previously I was managing the code bases and aws services manually, it was so pain, accidentally I discovered sam
and I introduced one template file for all the functions but build sam
was unable to find the code bases.
I specified CodeUri
where my codes points.
Now I can easily debug locally, build and deploy in one click. :-)
@jairoVera is life saver, I was banging on this problem for a week.
I have 13 serverless functions having go1.x and python run times and dependent to each others, previously I was managing the code bases and aws services manually, it was so pain, accidentally I discovered
sam
and I introduced one template file for all the functions but buildsam
was unable to find the code bases. Now I can easily debug locally, build and deploy in one click. :-)
So works or no? "but build sam was unable to find the code bases. Now I can easily debug locally, build and deploy in one click"
@jairoVera is life saver, I was banging on this problem for a week. I have 13 serverless functions having go1.x and python run times and dependent to each others, previously I was managing the code bases and aws services manually, it was so pain, accidentally I discovered
sam
and I introduced one template file for all the functions but buildsam
was unable to find the code bases. Now I can easily debug locally, build and deploy in one click. :-)So works or no? "but build sam was unable to find the code bases. Now I can easily debug locally, build and deploy in one click"
I already told.
"Now I can easily debug locally, build and deploy in one click. :-)"
I have a different question. How do I have 1 lambda function with different events (different http methods) I don't want to create several lambda functions for each method its not interesting and not efficient. I want to be able trigger that function with either with boolean logic inside the actual script or even better edit this serverless.yml file so I can do that from yaml rather than inside the script Please help me, will much appreciate that
It will generate two separate API gateways but the issue is (at least when generating a testing environment locally) if you create separate functions that have separate handler files both of the handler files will be written into the build folder for BOTH Lambda functions. I'm not sure if this would happen when deploying though.
Yes, it happens when you deploy it. You can see it in the AWS Lambda console. Were you (or someone) able to isolate the functions handlers when deploying?
In my case, when running sam build,
it executes the npm install for each function (Running NodejsNpmBuilder:NpmInstall
) and it takes a lot of time. This is the main issue I am trying to avoid.
I have a different question. How do I have 1 lambda function with different events (different http methods) I don't want to create several lambda functions for each method its not interesting and not efficient. I want to be able trigger that function with either with boolean logic inside the actual script or even better edit this serverless.yml file so I can do that from yaml rather than inside the script Please help me, will much appreciate that
This is what I do:
EcommerceFunction:
Type: AWS::Serverless::Function
Properties:
FunctionName: ecommerce-lambda
Handler: src/handlers/ecommerceHandler.handler
Runtime: nodejs12.x
Description: Products Lambda function.
Policies:
- AWSLambdaBasicExecutionRole
- AmazonDynamoDBFullAccess
Environment:
Variables:
LAMBDA_ENVIRONMENT: local
Events:
# Server
GetServiceRunningAPI:
Type: Api
Properties:
Path: /api/products-health-check
Method: GET
# Products
GetProductsAPI:
Type: Api
Properties:
Path: /api/products
Method: GET
GetSingleProductsAPI:
Type: Api
Properties:
Path: /api/products/{productId}
Method: GET
PostProductsAPI:
Type: Api
Properties:
Path: /api/products
Method: POST
PutProductsAPI:
Type: Api
Properties:
Path: /api/products/{productId}
Method: PUT
DeleteProductsAPI:
Type: Api
Properties:
Path: /api/products/{productId}
Method: DELETE
Hi, how does this work with java maven reactor build, is every module considered a lambda function? Is a deployment package generated from the root module contains all lambda on one single jar?
This is what I do:
... PutProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: PUT DeleteProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: DELETE
Only example I've been able to find where multiple http methods are specified from the same endpoint. Thank you!
This is what I do:
... PutProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: PUT DeleteProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: DELETE
Only example I've been able to find where multiple http methods are specified from the same endpoint. Thank you!
The million dollar question, how do you do it multiple paths and methods for HTTP APIs?
I have a different question. How do I have 1 lambda function with different events (different http methods) I don't want to create several lambda functions for each method its not interesting and not efficient. I want to be able trigger that function with either with boolean logic inside the actual script or even better edit this serverless.yml file so I can do that from yaml rather than inside the script Please help me, will much appreciate that
This is what I do:
EcommerceFunction: Type: AWS::Serverless::Function Properties: FunctionName: ecommerce-lambda Handler: src/handlers/ecommerceHandler.handler Runtime: nodejs12.x Description: Products Lambda function. Policies: - AWSLambdaBasicExecutionRole - AmazonDynamoDBFullAccess Environment: Variables: LAMBDA_ENVIRONMENT: local Events: # Server GetServiceRunningAPI: Type: Api Properties: Path: /api/products-health-check Method: GET # Products GetProductsAPI: Type: Api Properties: Path: /api/products Method: GET GetSingleProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: GET PostProductsAPI: Type: Api Properties: Path: /api/products Method: POST PutProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: PUT DeleteProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: DELETE
This is the way
one handler for all of the products API?
one handler for all of the products API?
Yes, I had that approach initially (one handler for all of the products API). Because on develop building several functions for each change was not possible. Right now I have moved 1 function per API endpoint.
Comment 1:
Right now (Feb 2021 - and a couple of versions before), sam build
has a --cached
and --parallel
parameters that helps to build faster
Comment 2:
We have moved to typescript with a webpack solution, so real hot-reload is achieved even without building (sam build
)
@sromano88-svc @estebansolo That looks so so nice!
one handler for all of the products API?
Yes, I had that approach initially (one handler for all of the products API). Because on develop building several functions for each change was not possible. Right now I have moved 1 function per API endpoint.
Comment 1: Right now (Feb 2021 - and a couple of versions before),
sam build
has a--cached
and--parallel
parameters that helps to build fasterComment 2: We have moved to typescript with a webpack solution, so real hot-reload is achieved even without building (
sam build
)
How do you do that ? What is your file-system structure ? I have webpack too and plan to use typescript
From https://github.com/aws-samples/aws-serverless-samfarm/issues/5#issuecomment-763156337
I have a different question. How do I have 1 lambda function with different events (different http methods) I don't want to create several lambda functions for each method its not interesting and not efficient. I want to be able trigger that function with either with boolean logic inside the actual script or even better edit this serverless.yml file so I can do that from yaml rather than inside the script Please help me, will much appreciate that
This is what I do:
EcommerceFunction: Type: AWS::Serverless::Function Properties: FunctionName: ecommerce-lambda Handler: src/handlers/ecommerceHandler.handler Runtime: nodejs12.x Description: Products Lambda function. Policies: - AWSLambdaBasicExecutionRole - AmazonDynamoDBFullAccess Environment: Variables: LAMBDA_ENVIRONMENT: local Events: # Server GetServiceRunningAPI: Type: Api Properties: Path: /api/products-health-check Method: GET # Products GetProductsAPI: Type: Api Properties: Path: /api/products Method: GET GetSingleProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: GET PostProductsAPI: Type: Api Properties: Path: /api/products Method: POST PutProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: PUT DeleteProductsAPI: Type: Api Properties: Path: /api/products/{productId} Method: DELETE
This is the way
If we need multiple languages on the same API path (we have a library which only exists in python, but our code is otherwise kotlin, and we need time to port the library to jvm), is there a way to do so?
Or would we be forced to use a separate API in this case?
(also in general I'm curious if there's a way to use the --use-container
flag with sam build
for heterogeneous runtimes without specifying a custom docker image)
@hughesjj This example shows how you can use the same lambda function (also language) with different endpoints.
If you want to use the same api with different endpoints and every of them with a different language, you can do it easily, check a previous comment where there are two functions with the same api, all you have to do is change the runtime and do what you need to do
@jairoVera's solution worked for me I just created two resource in same templete.yml file. here is my folder structure But node dependency will be heavy worried about optimization in build process.
Root Dir
api1 - api1 folder
index.js - handler file
node_modules/
api2 - api2 folder
index.js - handler file
node_modules/
template.yml file
** yml file **
Globals:
Function:
Timeout: 100
AutoPublishAlias: live
DeploymentPreference:
Enabled: true
Type: Canary10Percent5Minutes
Role: !Ref CodeDeployRole
Resources:
api1:
Type: AWS::Serverless::Function
Properties:
FunctionName: !Sub <Fn name>
Handler: index.handler
Runtime: nodejs12.x
CodeUri: api1/
Environment:
Variables:
NODE_ENV: ""
MONGO_URI: ""
Role:
Fn::GetAtt:
- LambdaExecutionRole
- Arn
Events:
GetEvent:
Type: Api
Properties:
Path: /
Method: get
api2:
Type: AWS::Serverless::Function
Properties:
FunctionName: !Sub <Fn Name>
Handler: index.handler
Runtime: nodejs12.x
CodeUri: api2/
Environment:
Variables:
NODE_ENV: ""
MONGO_URI: ""
Role:
Fn::GetAtt:
- LambdaExecutionRole
- Arn
Events:
GetEvent:
Type: Api
Properties:
Path: /get-user
Method: get