Supplier service should be repurposed (see blockchain use case) and the inventory service should serve the calls to add/remove/get inventory as was the case originally. This conforms to proper microservice arch more accurately re relation to pdb/schema and will allow us to use the various inventory implementation to show rest/sync communication as well as messaging.

If a UI password is specified that contains the ":" colon character then it is not possible to login.

Would be very nice to allow users to work from local machine rather than cloud shell once the initial workshop setup is complete.

The two main tasks here being...

1. either allow port of cloud shell env (ie instructions to copy "state" folder and source xyz to get state_get on path etc.) or allow local setting/override of DOCKER_REGISTRY etc.

2. get rid of sed usage as it's not OS portable (fails on mac).

Known anomalies:

• Free trial uses VM.Standard.E2.1, LiveLabs uses VM.Standard2.1. Ideally they would be the same. What to do?
• VM.Standard.E2.1 shapes became "End of orderability" in Feb 2021. We are still using them and they appear to work, but is there a risk they will stop working?
• Free trial scaling test runs approximately 40% faster on free trial than LiveLabs. Is this because of the difference in compute shape?
• My (Richard Exley) free trial account came with VM.Standard.E2.1 limit 6 - provisioned mid 2020. But this shape became "End of orderability" in Feb 2021. Do new free trial accounts have a different compute shape? Do we need to switch to a new shape? Does setup need to dynamically switch to an available shape?

https://github.com/oracle/microservices-datadriven/wiki/Sagas
Workshop that shows use of beta OSaga feature in DB with PL/SQL, Java, Apex Workflow.
WIP doc: https://paulparkinson.github.io/learning-library/developer-library/microservices-sagas/workshops/freetier

• #173
• #241

Test all Microservices to make sure we have transactional "all or nothing" behavior.

Enhance our Microservices to respond to shutdown signals and gracefully release all resources, for example DB connections.

Oracle supports several deployment models - multiple CDB, multiple PDB, multiple schema, single schema. Some require AQ propagation and some do not. It should to change the deployment model without having to change the Microservice code.

The problem is fixed by adding a filter to the oci_core_services data definition in core.tf:

data "oci_core_services" "services" { filter { name = "name" values = ["All .* Services In Oracle Services Network"] regex = true } }

Then referring to the first element in the oci_core_service_gateway and routing and security rules:

service_id = data.oci_core_services.services.services.0.id

destination = data.oci_core_services.services.services.0.cidr_block

Some ideas:

1. Try running in parallel
2. yum is slow for python on nodejs builds. Maybe download instead (as with inventory-go build)

• doc mock-up WIP is here: https://paulparkinson.github.io/learning-library/developer-library/microservices-with-converged-db/workshops/multicloud-freetier
• verifying Jaeger works in this env and perhaps adding kiali
• working on Azure now

For example: export DB_PASSWORD='Welcome12345;#!:'

Remove commented out code and add descriptive comments

The inventory service is

1. a headless service and
2. SQL/relational access only as far as the grabdish architecture.
   However, the inventory service also serves to demonstrate equivalent functionality in different languages and frameworks and therefore we should add.
3. A Rest endpoint (eg for report use case or some such)
4. Use both SQL and JSON (via JSON flex-field) to query and update inventory db.

Related reading...
https://blogs.oracle.com/sql/how-to-store-query-and-create-json-documents-in-oracle-database
https://docs.oracle.com/en/database/oracle/oracle-database/18/adjsn/function-JSON_TABLE.html#GUID-0172660F-CE29-4765-BF2C-C405BDE8369A
https://docs.oracle.com/en/database/oracle/oracle-database/12.2/objff/

Goals:

• Cover all lab content
• Cover all platforms (free tier, live labs)
• Hands free (lights out) with minimal prep
• Clear success or failure indication
• Setup, Test, Teardown

So it is ready when we do the scaling lab

Make teardown work correctly even if the setup was not completed successfully.

Microservices should not crash if the database is temporarily unavailable. Functions that do not require the database should remain available. Database connections should gracefully switchover to a surviving instance on instance shutdown with no application outage.

Lab 5 in data-driven ms workshop. Includes metrics, tracing, and logs for app and db tier in single-pane grafana view

Occasionally, although it appears that the setup has completed successfully, the frontend UI will hang and the login dialog is not displayed. This reproduced twice on May 25. It was found that the appropriate security rules had not been created in the grabdish VCN. This should have been done automatically when the frontend services was created but for some reason this was not done. There are two security rules required:

| Security List | Type    | Stateless | Destination/Source | Protocol | Source Port range | Destination Port Range |
|---------------|---------|-----------|--------------------|----------|-------------------|------------------------|
| svcLB         | Egress  | No        | 10.0.10.0/24       | IP       | All               | <PORT>                 |
| nodepool      | Ingress | No.       | 10.0.20.0/24       | IP       | All               | <PORT>                 |

The port is listed by calling kubectl get service frontend -n msdataworkshop, for example 30101 in this output:

NAME       TYPE           CLUSTER-IP     EXTERNAL-IP      PORT(S)         AGE
frontend   LoadBalancer   10.96.225.36   129.159.33.154   443:30101/TCP   77m

ie inventory-micronaut

need an ORM example as it is popular usage. micronaut data also runs on Helidon and supplier service is a good candidate regardless

This should be a continuing process but this issue is specifically for setting up and creating at least one test for each microservices to that they are easy to add.

I imagine this will be run as part of the regression suite where ./test.sh includes a call to /utils/unit-tests.sh which (similar to ./deploy.sh etc in the root dir) iterates over each of the mservices running their tests.

We may need to add -DskipTests to the usual build/package commands to avoid running tests during the workshop builds.

Just need tp decide where?

ie inventory-springboot

add link for 1.4 release
clarify ocid for username in console

Turn on autoscale features the k8s and ATP database level and show how resources scale up to handle a workload.

this should not actually be necessary and we can instead mount the files in the container at the expected locations at runtime rather than modify the WALLET_LOCATION eg.

Add support for the workshop to run in a Live Labs shared tenancy.

Step 7/10 : FROM mcr.microsoft.com/dotnet/aspnet:5.0
Trying to pull repository mcr.microsoft.com/dotnet/aspnet ...
Get "https://mcr.microsoft.com/v2/": net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)

Will see if I can repro

Show how a Microservice can be upgraded without impacting service availability

Show how queries can be executed across data stores using SQL (with domain specific extensions) and database links.