Skip to content

Latest commit

 

History

History
193 lines (138 loc) · 6.06 KB

README.md

File metadata and controls

193 lines (138 loc) · 6.06 KB
Raw

Ops Catalog Examples

There are a bunch of example configurations available in this project to run Ops Catalog with various extensions.

Running Ops Catalog Api

Ensure you have Docker/Nerdctl installed and execute the following:

git clone [email protected]:ops-catalog/examples.git
cd examples

Recipes

The recipes discussed here can be used for reference to try out a specific use case.

Recipe Name Goal
simple run a basic catalog API with static data
minimal run ui,catalog and postgres
ssl run ui, catalog and postgres. Postgres TLS is enabled.
edge run ui and multiple catalog instances. One acts as aggregator and another as edge
discovery finds resources in targets like databases, airflow, k8s, messaging servers
fulfillment provisions resources against databases, message servers

A run.sh file is provided to make it easier to run various docker compose files. Usage documentation for run.sh is shared below.

$ ./run.sh                                                                                                                 i
a recipe can be run like this:

  ./run.sh recipe <name>
  ./run.sh down

Recipe command runs one of the available recipes.

name:
-----
simple      - a catalog server with sample catalog items loaded from ./datasets/catalog/my-account
minimal     - a basic setup with catalog, ui, postgres and some seed data
ssl         - a basic setup with catalog, ui and postgres with TLS
edge        - aggregator catalog discovers data from edge catalog
discovery   - performs discovery against few targets like postgres, cassandra, airflow, k8s, kafka and presents in UI
fulfillment - provisions resources in catalog which are not yet in target servers

Down command shuts down the running containers.

Recipe: simple

This is probably the simplest use case. We are mounting catalog items into a folder and serving it via catalog instance. This minimilastic config can be queried and filtered through a HTTP API call.

To run this recipe, invoke

./run.sh recipe simple

Docker containers started with ./run.sh command can be stopped using ./run.sh down

Since this step only requires a single docker container to run, it can also be invoked directly like so:

docker run \
-v $(pwd)/datasets:/opt/app/datasets \
-v $(pwd)/recipes/simple:/opt/app/config \
-e CONFIG_FILE=/opt/app/config/config.yaml -p 8080:8080 \
-it opscatalog/catalog:latest

Test Access

Load the following link to view catalog item:

Link Description
http://localhost:8080/api/catalog Catalog Listing
http://localhost:8080/api/catalog?kind=Resource List Resources Only
http://localhost:8080/api/catalog?kind=Component List Components Only

Recipe: minimal

If you have resource constraint, you can run selected profile as well and accordingly update engines list in setup/containers/ops-catalog/conf/discovery.yaml or fulfillment.yaml

We can run the command like below to bring up ops catalog API, a stand-in for objects in a Kubernetes cluster, a postgres instance with two databases and a basic UI.

./run.sh recipe minimal

To test discovery and fulfillment, create a new schema against the running postgres.

docker exec -it postgres psql -Upostgres -dservicing 
servicing=# CREATE SCHEMA IF NOT EXISTS refdata;
servicing=# \q

Also drop a new file into the mix

cat > datasets/my-account/merchant-schema.yaml <<EOF
apiVersion: "v1"
kind: Resource
class: Schema
metadata:
  name: "merchants"
  description: "Merchant Tables are hosted in this schema"
  license: "private"
dependencies:
  upstream: []
  providedBy: postgres.pg-2
  triggers: []
classification:
    tag: ["transaction", "customer"]
    domain: "transaction"
    team: "transaction"
    capability: "onlinebanking"
EOF

Once the discovery and fulfillment loop is complete, there should be two new items in the catalog. The schema called merchant should be visible in postgres as well.

docker exec -it postgres psql -Upostgres -dpreferences
preferences=# select catalog_name, schema_name from information_schema.schemata;
preferences=# \q

Check the catalog entry via api calls. The default refresh frequency is served data is set at 5 minutes which you can change by updating ops-catalog/conf/*.yaml

http://localhost:8080/api/catalog?name=merchants
http://localhost:8080/api/catalog?name=servicing.refdata

If you list files under datasets/discovered-items you should also see few new folders depending on what discovery engines were enabled. In the case of minimal profile, you will see k8s and postgres folders populated with catalog items.

UI

Ops Catalog can provide data to your existing backstage instance. Navigate to http://localhost:7007/catalog to see Ops Catalog objects within backstage UI.

Tearing Down

Each Recipe run can be cleaned up by running ./run.sh down

Recipe: ssl

To run with SSL enabled, set SSL_STATE to on in your .dockerenv or one of the env files. A .ssl file is provided for reference and this setup can be run with SSL by executing:

./run.sh recipe ssl

Recipe: edge

A number of catalog instances can be run in federated mode. This allows for local teams to run their own edge catalog for their unique workflows. A more global catalog instance can then act as aggregator to compose data from all catalogs.

You can simulate this by running the following:

./run.sh recipe edge

Recipe: discovery

This compose file also runs a standalone kafka, postgres and cassandra and seeds them with initial objects so they can be collected by the discovery module.

./run.sh recipe discovery

Recipe: fulfillment

The compose file is somewhat similar to discovery as it is now writing back to the targets.

./run.sh recipe fulfillment

Refer to the documentation for details.

Cleanup Activity

To cleanup what we did just then, run the following command to remove all running containers associated with this project.

./run.sh down

or

docker compose down