you need to take the artifact produced and "get it" to your server ... e.g. using a mounted volume in your k8s

We'll update the documentation to make this clearer!

is there an alternative method for adding aspects at runtime via an API?

or all methods require placing the PDL onto the GMS server

@early-lamp-41924 might be able to give you some advice for how to set up a dynamic mount like this

so first of all, by default it reads models from /etc/datahub/plugins/models https://github.com/linkedin/datahub/blob/master/metadata-service/factories/src/main/resources/application.yml#L44 but you can update this behavior with an env variable

that being said we need to mount a local jar file into the k8s pod. and that is done through volumes and volume mounts

First you need to push the jar into a configmap refer to this command

kubectl create configmap custom-model --from-file=<<path-to-jar-file>> -n <<namespace>>

Then you need to set the volumes for the gms deployment (refer to how jmx exporter configmap is added here https://github.com/acryldata/datahub-helm/blob/master/charts/datahub/subcharts/datahub-gms/templates/deployment.yaml#L40) This just tells the gms deploy in advance that we will be pulling this configmap in. You can do this by setting the datahub-gms.extraVolumes which gets appended to the deployment without having to change the helm chart.

Finally you need to mount the volume into the container’s local directory by setting volumeMounts. Refer to how the kafka certs are mounted onto a local path here https://github.com/acryldata/datahub-helm/blob/master/charts/datahub/subcharts/datahub-gms/templates/deployment.yaml#L182 You can do this by setting the datahub-gms.extraVolumeMounts

at the end your values.yaml should have something like:

datahub-gms:
  ...
  extraVolumes:
    - name: custom-model
      configMap:
        name: custom-model ## should match configmap name above
  extraVolumeMounts:
    - name: custom-model-dir
      mountPath: /etc/datahub/plugins/models

Let us know how this goes

I uploaded the jar from build/libs

metadata-models-custom-data-template-0.0.8.jar

Archive: metadata-models-custom-data-template-0.0.8.jar Length Date Time Name --------- ---------- ----- ---- 0 2022-01-28 15:45 META-INF/ 25 2022-01-28 15:45 META-INF/MANIFEST.MF 0 2022-01-28 15:45 com/ 0 2022-01-28 15:45 com/factset/ 0 2022-01-28 15:45 com/factset/bmd/ 1496 2022-01-28 15:45 com/factset/bmd/BusinessMetadata$Fields.class 8207 2022-01-28 15:45 com/factset/bmd/BusinessMetadata.class 0 2022-01-28 15:45 pegasus/ 0 2022-01-28 15:45 pegasus/com/ 0 2022-01-28 15:45 pegasus/com/factset/ 0 2022-01-28 15:45 pegasus/com/factset/bmd/ 312 2022-01-28 15:45 pegasus/com/factset/bmd/BusinessMetadata.pdl 0 2022-01-28 15:45 legacyPegasusSchemas/ 0 2022-01-28 15:45 legacyPegasusSchemas/com/ 0 2022-01-28 15:45 legacyPegasusSchemas/com/factset/ 0 2022-01-28 15:45 legacyPegasusSchemas/com/factset/bmd/ 607 2022-01-28 15:45 legacyPegasusSchemas/com/factset/bmd/BusinessMetadata.pdsc --------- ------- 10647 17 files

then on the pod: /etc/datahub/plugins/models $ ls metadata-models-custom-data-template-0.0.8.jar

Do I need to mount the zip file and then add a command to extract at container startup?

if you run the gradle task

modelArtifact

in the custom-metadata-model module

you will find a zip file generated under

build/dist

something like

metadata-models-custom-0.0.1.zip

e.g.

../gradlew modelArtifact -PprojVersion=0.0.2

will generate a zip file with the

0.0.2

version

this zip file contains everything you need

you just need to drop it (unzipped) under the

/etc/plugins/models/<registry-name>/<registry-version>

directory

just upgraded

ok it works now 🙂

thank you

should the UI also update after a deployment of an artifact

namespace com.factset.bmd @Aspect = { "name": "businessMetadata", "autoRender": true, "renderSpec": { "displayType": "properties", "displayName": "Business Metadata" } } record BusinessMetadata { ... }

So I can deploy the model

then use the sample script to add a record

datahub put --urn "urnliglossaryTerm:(Test.test)" --aspect businessMetadata --aspect-data data/bmd.json

but when i navigate to the glossary term Test.test, I don't see anything in the UI for the businessAspect I created

it should be available soon as Glossary terms are getting moved over to the new UI rendering today

for use with auto rendering

also, when they are moved over to the the UI rendering, will that enable it, or maybe more work after?

We should also document which entities will work with auto-render aspects, will do that soon.