Manage workload merge behavior

This topic tells you how to manage the workload update behavior with the Tanzu Apps CLI --update-strategy flag.

When updating a workload from a file, manage the workload update behavior with the --update-strategy flag. There are two possible values: merge or replace. The default value is merge.

merge

If the --file workload.yaml deletes an existing on-cluster property or value, it is not removed from the on-cluster definition. If the --file workload.yaml includes a new property or value, it is added to the on-cluster workload property value. If the --file workload.yaml updates an existing value for a property, it is updated on the on-cluster definition.

replace

The on-cluster workload is updated to exactly what is specified in the --file workload.yaml definition.

The current default merge strategy intents to prevent unintentional deletions of critical properties from existing workloads.

Note

The default value for the --update-strategy flag will change from merge to replace in Tanzu Application Platform v1.7.0.

Examples of the outcomes of both the merge and replace values:

# Export workload if there is no previous yaml definition
tanzu apps workload get spring-petclinic --export > spring-petclinic.yaml

# modify the workload definition
vi rmq-sample-app.yaml
---
apiVersion: carto.run/v1alpha1
kind: Workload
metadata:
  name: spring-petclinic
  labels:
    app.kubernetes.io/part-of: spring-petclinic
    apps.tanzu.vmware.com/workload-type: web
spec:
  resources:
    requests:
      memory: 1Gi
    limits:           # delete this line
      memory: 1Gi     # delete this line
      cpu: 500m       # delete this line
  source:
    git:
      url: https://github.com/sample-accelerators/spring-petclinic
      ref:
        tag: tap-1.1

After saving the file, to verify how both of the update strategy options behave, run:

tanzu apps workload apply -f ./spring-petclinic.yaml --update-strategy merge # if flag is not specified, merge is taken as default

This produces the following output:

❗ WARNING: Configuration file update strategy is changing. By default, provided configuration files
will replace rather than merge existing configuration. The change will take place in the January 2024
Tanzu Application Platform release (use "--update-strategy" to control strategy explicitly).

Workload is unchanged, skipping update

By contrast, use replace as follows:

tanzu apps workload apply -f ./spring-petclinic.yaml --update-strategy replace

This produces the following output:

❗ WARNING: Configuration file update strategy is changing. By default, provided configuration files
will replace rather than merge existing configuration. The change will take place in the January 2024
Tanzu Application Platform release (use "--update-strategy" to control strategy explicitly).

🔎 Update workload:
...
  8,  8   |  name: spring-petclinic
  9,  9   |  namespace: default
 10, 10   |spec:
 11, 11   |  resources:
 12     - |    limits:
 13     - |      cpu: 500m
 14     - |      memory: 1Gi
 15, 12   |    requests:
 16, 13   |      memory: 1Gi
 17, 14   |  source:
 18, 15   |    git:
...
❓ Really update the workload "spring-petclinic"? [yN]:

The lines that were deleted in the yaml file are deleted as well in the workload running in the cluster. The only text boxes that remain exactly as they were created are the system populated metadata text boxes; resourceVersion, uuid, generation, creationTimestamp, and deletionTimestamp.