Skip to content
This repository has been archived by the owner on Nov 17, 2021. It is now read-only.

Latest commit

 

History

History

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

The BRAIN component

Contents

Overview

This section relates to BRAIN component, the service that coordinates the event messages, subscribes to the "salesorder/created" topic, and publishes event messages to the "Internal/Charityfund/Increased" topic, and is represented by the "BRAIN" block in the whiteboard diagram.

The Brain in context

The BRAIN component is a basic CAP application with two of the three layers in use. In effect, a "service" more than an application:

Layer Description
app Unused
srv A single service teched is defined exposing an entity CharityEntry. Custom JavaScript code for managing the BRAIN's operations and activities
db An entity CharityEntry is defined with a SoldToParty property as key, and a counter property. This entity is defined within the namespace charity

The service, once deployed, does not require any human intervention to function. Processing follows a sequence of the following activities, each time an event published to the "salesorder/created" topic on the message bus is received; each activity is denoted by a "level" number (1 through 4):

  1. Log the message details
  2. Retrieve sales order header details from the OData service API_SALES_ORDER_SRV proxied by the SANDBOX component
  3. Request a conversion of the total net amount of the sales order to the equivalent in charity fund credits, by calling the CONVERTER component*
  4. Publish a new event to the "Internal/Charityfund/Increased" topic*

*This is as long as the sold-to party is one that hasn't already been processed 10 times before

Controlling the processing

To aid testing and gradual component deployment (getting all components up and running and connected), an enhancement has been made since the Developer Keynote presentation to allow the control of these activities using an environment variable BRAIN_LEVEL.

Setting this to 0 will mean that none of the above activities are carried out. Setting this to a value equivalent to one of the activity numbers (i.e. 1, 2, 3 or 4) will mean that activities up to and including that number will be carried out. The default value (if none is set explicitly) is 1, meaning the received event message will be logged, and that's all.

Remote services defined and used

In carrying out the activities listed above, the CAP service consumes the following services which are defined in package.json. Some of these employ destinations defined in the cloud (on CF - see below):

Name Kind Details
messaging enterprise-messaging Connection to the message bus
db sqlite Local persistence
S4SalesOrders odata Connection to the API Hub sandbox (SAP S/4HANA Cloud mock system) OData service. Destination apihub_mock_salesorders
ConversionService rest Connection to the Converter (Go microservice) conversion service. Destination charityfund_converter

Setup required

You'll need to set a few things up in preparation for getting this component up and running.

Assumptions

These instructions assume you've cloned your forked copy of this repository, as described in the Download and installation section of the main repository README. The GitHub username in the examples in this README is 'qmacro' - you should replace this with your own GitHub username.

It also assumes you've opted to use the SAP Business Application Studio (App Studio) for your development environment and have followed the setup instructions in Using the SAP Business Application Studio. Prompts shown in examples in this README will reflect the prompt style in the terminal of an App Studio Dev Space (showing the username, current directory, and so on).

Finally it also assumes you've already set up the message bus, an instance of the SAP Enterprise Messaging service, following the instructions in Message bus setup.

Destinations

As mentioned above, there are some destinations at play here. Two, in fact, pointing to:

  • the S4SalesOrders endpoint, required for BRAIN_LEVEL 2
  • the ConversionService endpoint, required for BRAIN_LEVEL 3

If you're just starting out, and only want to get the BRAIN component up and running at BRAIN_LEVEL 1 for now, you can leave the destinations setup until you're ready. Otherwise, set these destinations up now, following the Destinations setup instructions.

Service instances

In addition to an instance of the SAP Enterprise Messaging service, the BRAIN component requires access to instances of another couple of services:

  • the Destination service
  • the Authorization & Trust Management service (otherwise known as XSUAA)

These services are to enable the reading of those two destination definitions (for the S4SalesOrders and ConversionService endpoints) that you've just created in the previous section.

Set up these services now following the Service instances setup instructions.

Local service setup

The best way to get this component up and running is to start locally. So now is a good point to set things up for local execution. This is a CAP based service, which relies on certain NPM modules (see the dependencies and devDependencies nodes in package.json and a local SQLite-powered persistence layer (see the cds -> requires -> db node in the same file).

First, ensure you're in this (cap/brain/) directory:

user: teched2020-developer-keynote $ cd cap/brain/
user: brain $ pwd
/home/user/projects/teched2020-developer-keynote/cap/brain

Next, get the modules installed by running npm install. This is the sort of thing you should see:

user: brain $ npm install

> @sap/[email protected] install /private/tmp/teched2020-developer-keynote/cap/brain/node_modules/@sap/hana-client
> node checkbuild.js

> [email protected] install /private/tmp/teched2020-developer-keynote/cap/brain/node_modules/sqlite3
> node-pre-gyp install --fallback-to-build

node-pre-gyp WARN Using needle for node-pre-gyp https download
[sqlite3] Success: "/private/tmp/teched2020-developer-keynote/cap/brain/node_modules/sqlite3/lib/binding/node-v72-darwin-x64/node_sqlite3.node" is installed via remote

> @sap-cloud-sdk/[email protected] postinstall /private/tmp/teched2020-developer-keynote/cap/brain/node_modules/@sap-cloud-sdk/core
> node usage-analytics.js

added 224 packages from 153 contributors and audited 224 packages in 6.956s

4 packages are looking for funding
  run `npm fund` for details

found 0 vulnerabilities

Now run cds deploy to cause the persistence layer artifact (the SQLite database file) to be summoned into existence (note that there are a few test records in the CSV file):

user: brain $ cds deploy
 > filling charity.CharityEntry from db/csv/charity-CharityEntry.csv
/> successfully deployed to ./brain.db

At this stage you're ready to embark upon running this component locally.

Running it

In a similar way to the SANDBOX component, you can get this component up and running at different levels - in this case we'll get it running locally and on Kyma.

Locally

It's straightforward to run CAP applications and services locally, but when they consume cloud-based services, connection and credential information is required, and for local execution, this information is traditionally stored in a file called default-env.json.

Because of what this contains, it is not normally included in any repository for security reasons, so you should generate this yourself now.

At this point, you can start the service running locally with cds run, shown here with typical output (with some lines removed for readability):

user: brain $ cds run

[cds] - model loaded from 3 file(s):

  db/schema.cds
  srv/service.cds
  srv/external/API_SALES_ORDER_SRV.csn

[cds] - connect to db > sqlite { database: 'brain.db' }
[cds] - connect to messaging > enterprise-messaging {
  management: [
    {
      oa2: {
        clientid: 'sb-clone-xbem-service-broker-aec3bfac91f84d61841ef28efb7fa235-clone!b68527|xbem-service-broker-!b2436'
      },
      uri: 'https://enterprise-messaging-hub-backend.cfapps.eu10.hana.ondemand.com'
    }
  ],
  messaging: [
    {
      broker: { type: 'sapmgw' },
      oa2: {
        client: 'sb-clone-xbem-service-broker-aec3bfac91f84d61841ef28efb7fa235-clone!b68527|xbem-service-broker-!b2436'
      },
      protocol: [ 'amqp10ws' ],
      uri: 'wss://enterprise-messaging-messaging-gateway.cfapps.eu10.hana.ondemand.com/protocols/amqp10ws'
    },
    {
      broker: { type: 'sapmgw' },
      oa2: {
        clientid: 'sb-clone-xbem-service-broker-aec3bfac91f84d61841ef28efb7fa235-clone!b68527|xbem-service-broker-!b2436'
      },
      protocol: [ 'mqtt311ws' ],
      uri: 'wss://enterprise-messaging-messaging-gateway.cfapps.eu10.hana.ondemand.com/protocols/mqtt311ws'
    },
    {
      broker: { type: 'saprestmgw' },
      oa2: {
        clientid: 'sb-clone-xbem-service-broker-aec3bfac91f84d61841ef28efb7fa235-clone!b68527|xbem-service-broker-!b2436'
      },
      protocol: [ 'httprest' ],
      uri: 'https://enterprise-messaging-pubsub.cfapps.eu10.hana.ondemand.com'
    }
  ],
  serviceinstanceid: 'aec3bfac-91f8-4d61-841e-f28efb7fa235',
  xsappname: 'clone-xbem-service-broker-aec3bfac91f84d61841ef28efb7fa235-clone!b68527|xbem-service-broker-!b2436'
}
BRAIN_LEVEL set to 1
[cds] - Put queue { queue: 'CAP/0000' }
[cds] - serving API_SALES_ORDER_SRV { at: '/api-sales-order-srv' }
[cds] - serving teched { at: '/teched', impl: 'srv/service.js' }

[cds] - launched in: 1258.575ms
[cds] - server listening on { url: 'http://localhost:4004' }
[ terminate with ^C ]

[cds] - Add subscription { topic: 'salesorder/created', queue: 'CAP/0000' }

Before we start to examine the output, if you're in Dev Space from your App Studio based development environment, the fact that a process has started listening on host localhost port 4004 will have caused App Studio to prompt you with a message offering you the possibility of exposing this so that you can access the service from your browser:

port message

Select the "Expose and Open" button (you may be subsequently prompted to add a description). Once you've seen the contents of the web page, flip back to your browser tab where you have your Dev Space, to look at the output of cds run.

In that output, observe how the CAP messaging support automatically connects to the message bus (the instance of the SAP Enterprise Messaging service) and, in order to subscribe to the "salesorder/created" topic, creates a queue "CAP/0000" and a queue subscription, connecting that "CAP/0000" queue to the "salesorder/created" topic:

[cds] - Put queue { queue: 'CAP/0000' }
...
[cds] - Add subscription { topic: 'salesorder/created', queue: 'CAP/0000' }

See the Diving into messaging on SAP Cloud Platform series on the SAP Developers YouTube channel for explainations of how queues, topics and queue subscriptions work, and plenty more besides.

Observe also this message:

BRAIN_LEVEL set to 1

This is directly related to the activity level described earlier in the Controlling the process section, and the value of 1 (for logging the message details only) is the default value.

Testing BRAIN_LEVEL 1

At this point, you should leave this CAP service running, and (say, in a new terminal window) jump over to your EMITTER component, set that up (if you haven't got it set up already) and emit a "salesorder/created" event message. Look in particular at the Usage section for hints on how to do this. Emit an event message for a sales order (e.g. 1) - the invocation and output should look something like this:

user: event $ ./emit 1
2020-12-07 13:48:59 Publishing sales order created event for 1
2020-12-07 13:48:59 Publish message to topic salesorder%2Fcreated

More importantly, if you look back at the log output of your BRAIN component, you should see some extra log output, similar to this:

Message received {"_events":{},"_eventsCount":0,"_":{"event":"salesorder/created","data":{"SalesOrder":"1"},"headers":{"type":"sap.s4.beh.salesorder.v1.SalesOrder.Created.v1","specversion":"1.0","source":"/default/sap.s4.beh/DEVCLNT001","id":"ABFAF2F3-931F-49A8-86E8-876C295D9FAD","time":"2020-12-07T13:48:59Z","datacontenttype":"application/json"},"inbound":true},"event":"salesorder/created","data":{"SalesOrder":"1"},"headers":{"type":"sap.s4.beh.salesorder.v1.SalesOrder.Created.v1","specversion":"1.0","source":"/default/sap.s4.beh/DEVCLNT001","id":"ABFAF2F3-931F-49A8-86E8-876C295D9FAD","time":"2020-12-07T13:48:59Z","datacontenttype":"application/json"},"inbound":true}

This is the event message that the CAP service received from the message bus, because of its subscription to the "salesorder/created" topic. If we strip away the "Message received" text, it's JSON, and neatly formatted, we have:

{
  "_events": {},
  "_eventsCount": 0,
  "_": {
    "event": "salesorder/created",
    "data": {
      "SalesOrder": "1"
    },
    "headers": {
      "type": "sap.s4.beh.salesorder.v1.SalesOrder.Created.v1",
      "specversion": "1.0",
      "source": "/default/sap.s4.beh/DEVCLNT001",
      "id": "ABFAF2F3-931F-49A8-86E8-876C295D9FAD",
      "time": "2020-12-07T13:48:59Z",
      "datacontenttype": "application/json"
    },
    "inbound": true
  },
  "event": "salesorder/created",
  "data": {
    "SalesOrder": "1"
  },
  "headers": {
    "type": "sap.s4.beh.salesorder.v1.SalesOrder.Created.v1",
    "specversion": "1.0",
    "source": "/default/sap.s4.beh/DEVCLNT001",
    "id": "ABFAF2F3-931F-49A8-86E8-876C295D9FAD",
    "time": "2020-12-07T13:48:59Z",
    "datacontenttype": "application/json"
  },
  "inbound": true
}

Nice!

Depending on how far you've got with the setup of the other components in this repository, specifically those two that this component interact with - the SANDBOX and the CONVERTER components - you may want to set the value for the BRAIN_LEVEL accordingly.

Testing BRAIN_LEVEL 2

For example, if you've got the SANDBOX component set up (including a destination for it), you can increase the BRAIN_LEVEL to 2, to have the sales order header details retrieved from the OData service API_SALES_ORDER_SRV proxied by the that component.

A recent update to the libraries in CAP that provide this level of support means that if an entity in the imported OData service is prefixed with a namespace, then if the service definition has the same name as the namespace (in this case API_SALES_ORDER_SRV) then it will automatically try that prefix in building the query. If the service definition name is different (in this case, it is, being S4SalesOrders) then the namespace needs to be specified explicitly. This means that the SELECT target is now SELECT.one('API.SALES_ORDER_SRV.A_SalesOrder') rather than just SELECT.one('A_SalesOrder').

This is how you'd do that - basically you can specify the value for BRAIN_LEVEL, while restarting the service, this time giving an explicit value for the variable, like this:

user: brain $ BRAIN_LEVEL=2 cds run

In the other terminal window, emitting another event message in the same way (./emit 1) should result in not only the logging of the event message received, but also the results of the sales order information retrieval described (in the overview) as what happens at BRAIN_LEVEL 2.

This is the sort of thing you should see (note the log output shown here starts with BRAIN_LEVEL set to 2, but omits the message logging output from BRAIN_LEVEL 1):

BRAIN_LEVEL set to 2
...
SalesOrder number is 1
[cds] - connect to S4SalesOrders > odata {
  destination: 'apihub_mock_salesorders',
  path: '/sap/opu/odata/sap/API_SALES_ORDER_SRV'
}
SalesOrder details retrieved {"SalesOrder":"1","SalesOrganization":"1710","SoldToParty":"17100001","CreationDate":"/Date(1471392000000)/","TotalNetAmount":"52.65"}

Observe that this time, not only is the event message logged ("Message received { ... }") but also a connection is made to the S4SalesOrders endpoint and header data is retrieved for the sales order number sent in the event message (1). This header data is in JSON format, and pretty-printed, looks like this:

{
  "SalesOrder": "1",
  "SalesOrganization": "1710",
  "SoldToParty": "17100001",
  "CreationDate": "/Date(1471392000000)/",
  "TotalNetAmount": "52.65"
}

Testing BRAIN_LEVEL 3

Once you also have the CONVERTER component up and active in the Kyma runtime in your trial subaccount (along with the corresponding destination for it), you can move up to BRAIN_LEVEL 3 to include the conversion from the total net amount of the sales order to the credit amount for the charity fund.

Restart the service, this time specifying 3 as the BRAIN_LEVEL value:

user: brain $ BRAIN_LEVEL=3 cds run

As before, emit another event message in the other terminal window, and check the log output from this CAP service. In addition to the messages we've already seen for BRAIN_LEVEL 1 and 2, you should now also see something like this:

BRAIN_LEVEL set to 3
...
[cds] - connect to ConversionService > rest { destination: 'charityfund_converter' }
Conversion result is {"Credits":7.9}

This shows that the CAP service successfully connected to the RESTful endpoint of the CONVERTER component and retrieved the charity fund credit amount equivalent for the sales order's total net amount.

Testing BRAIN_LEVEL 4

Finally you can test BRAIN_LEVEL 4, which causes an event message to be created and published to the message bus, specifically on the topic "Internal/Charityfund/Increased". Because there's no further service upon which the CAP service relies for this, you can go ahead straight away after BRAIN_LEVEL 3 and test it:

user: brain $ BRAIN_LEVEL=4 cds run

After emitting one more event message in the other terminal window, you should see extra log messages for this new BRAIN_LEVEL 4, something like this:

BRAIN_LEVEL set to 4
...
Payload for Internal/Charityfund/Increased topic created {"data":{"specversion":"1.0","type":"z.internal.charityfund.increased.v1","datacontenttype":"application/json","id":"9d25d3ed-76de-4d68-9bf9-f333710206d4","time":"2020-12-07T18:55:14.904Z","source":"/default/cap.brain/C02CH7L4MD6T","data":{"salesorder":"1","custid":"17100001","creationdate":"2020-12-08","credits":"7.9","salesorg":"1710"}}}
Published event to Internal/Charityfund/Increased

This represents the publishing of an event message that the CHARITY component is subscribed to, and the success of getting to this level in the CAP service testing marks the end of local execution testing.

Setting BRAIN_LEVEL permanently

Once you've reached this stage, you should set the BRAIN_LEVEL value permanently in the package.json-based start script definition. Edit package.json and add BRAIN_LEVEL=4 before the cds run for the "start" script, like this (just like you've been doing during testing):

{

  "scripts": {
    "start": "BRAIN_LEVEL=4 cds run",
    "hana": "cds deploy --to hana:brain --auto-undeploy",
    "build": "cds build/all --clean"
  },

}

On SAP Cloud Platform - Kyma runtime

Well done to making it this far in the README!

Now we've successfully got the service running locally, we'll go directly to a deployment to the Kyma runtime. (If you're interested in how you might also run a service on CF, see how we do it for the SANDBOX component.)

This BRAIN component has something in common with the SANDBOX and CONVERTER components ... and that is that the Kyma runtime was where each of them was running in the actual Developer Keynote. And the process for getting the component up and running in the Kyma runtime is the same each time.

Build & publish, secrets and deployment

Because the process is common, you can find it described in a separate (and therefore shared) document - Getting a component up and running on the Kyma runtime. Head on over to that document now, and follow the instructions, bearing in mind that you're building the BRAIN component. Come back here when you're done, because there's an extra step required for this BRAIN component, which we can do when you return.

Create and deploy a credentials config map

OK. Welcome back. In addition to the steps described in Getting a component up and running on the Kyma runtime, there's one more step for this BRAIN component, too:

  • create and deploy a credentials config map containing the access credentials for the services that the brain connects to

Unlike the SANDBOX and CONVERTER components, this component connects to and consumes various services, as you know. The credentials to make this possible have been in the default-env.json file, but we shouldn't include those credentials in any app image that is to be deployed. There are a couple of reasons that come to mind immediately: we should always avoid exposing such information in images especially when publishing them to a public registry, and also, we want to be able to manage the lifecycle of credentials independent of the app or service that uses them - otherwise we'd end up having to rebuild the app image each time.

This is where the deployment to Kyma differs from that of the SANDBOX component; this is the extra step mentioned above. If you take a look in the deployment.yaml file in this directory, you'll see a section like this:

envFrom:
  - configMapRef:
      name: appconfigcap

This refers to the config map that we're about to create in this step, containing the same VCAP_SERVICES environment variable based access credentials as earlier.

You'll need a default-env.json file. If you've already run the CAP brain service locally, you will have it. If not, follow the instructions in Generating a default-env.json file before continuing.

You can create the config map and deploy it in a single step, again using the k script. Here's an example invocation:

user: brain $ ./k configmap
Creating and deploying config map to k8s
configmap/appconfigcap configured

Check the component is up and running

Well done! At this stage, you should have the BRAIN component deployed to and running in the Kyma runtime in your SAP Cloud Platform trial account. If you check the Deployments in the Kyma console you might see something similar to this, where this BRAIN component (package name brain) is deployed, along with others (s4mock and calc-service in this example):

The CAP brain service deployed to the Kyma runtime