oc / kubernetes: deployment returns "unrecognized type: string" while adding environment variables - kubernetes

Question

We are deploying with Ansible scripts to Openshift 3 using oc apply. When we change template to add more environment variables, we receive a very vague error: "unrecognized type: string" and status code 500.
Setting --loglevel 10 leads to no more details:
$ /usr/local/bin/oc_v3.11.715 apply -f \"/tmp/ansible.YtEqVm_deploy/app.yml.json\" -n test-env --loglevel 10 2&> log.log
(several GET to get secret, deploymentconfigs, etc.)
...
I0127 11:49:05.455217 605 request.go:897] Request Body: {xxxxxxxx}
I0127 11:49:05.455280 605 round_trippers.go:386] curl -k -v -XPATCH -H "User-Agent: oc_v3.11.715/v1.11.0+d4cacc0 (linux/amd64) kubernetes/d4cacc0" -H "Authorization: Bearer xxxxxx" -H "Accept: application/json" -H "Content-Type: application/strategic-merge-patch+json" 'https://test-env:8443/apis/apps.openshift.io/v1/namespaces/test-app/deploymentconfigs/app'
I0127 11:49:05.466278 605 round_trippers.go:405] PATCH https://test-env:8443/apis/apps.openshift.io/v1/namespaces/test-env-app/deploymentconfigs/app 500 Internal Server Error in 10 milliseconds
I0127 11:49:05.466287 605 round_trippers.go:411] Response Headers:
I0127 11:49:05.466291 605 round_trippers.go:414] Content-Length: 118
I0127 11:49:05.466294 605 round_trippers.go:414] Date: Fri, 27 Jan 2023 09:49:05 GMT
I0127 11:49:05.466297 605 round_trippers.go:414] Audit-Id: 1d3f3398-14fc-4bfa-854b-6faf9b105680
I0127 11:49:05.466302 605 round_trippers.go:414] Cache-Control: no-store
I0127 11:49:05.466307 605 round_trippers.go:414] Content-Type: application/json
I0127 11:49:05.466321 605 request.go:897] Response Body: {"kind":"Status","apiVersion":"v1","metadata":{},"status":"Failure","message":"unrecognized type: string","code":500}
I0127 11:49:05.466603 605 helpers.go:201] server response object: [{
"kind": "Status",
"apiVersion": "v1",
"metadata": {},
"status": "Failure",
"message": "unrecognized type: string",
"code": 500
}]
F0127 11:49:05.466618 605 helpers.go:119] Error from server: unrecognized type: string
The request body is like:
{
"metadata": {
"annotations": {
"kubectl.kubernetes.io/last-applied-configuration": "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
}
},
"spec": {
"template": {
"spec": {
"$setElementOrder/containers": [{
"name": "app"
}],
"containers": [{
"$setElementOrder/env": [{
"name": "OLD_VAR_1"
}, {
"name": "OLD_VAR_2"
}, {
"name": "OLD_VAR_3"
}, {
"name": "OLD_VAR_4"
}, {
"name": "NEW_VAR_1"
}, {
"name": "NEW_VAR_2"
}, {
"name": "NEW_VAR_3"
}],
"dnsPolicy": "ClusterFirst",
"env": [{
"name": "OLD_VAR_4",
"value": false
}, {
"name": "NEW_VAR_1",
"value": 10
}, {
"name": "NEW_VAR_2",
"value": 20
}, {
"name": "NEW_VAR_3",
"value": 6
}],
"name": "app",
"restartPolicy": "Always",
"terminationGracePeriodSeconds": 300
}]
}
}
}
}
OLD_VAR_x are old environment variables; we want to add NEW_VAR_[1-3]. Notice strangely that not all old vars are in env, only OLD_VAR_4, but all new vars are in env.
This also happens when we use oc patch with the same request body. Same error response.
What is wrong?
A workaround is first, deployment, fail, and add new vars in Openshift manually, and deploy in Openshift webconsole on top of the last, failed deployment. It works.

Answer 1

Solved by quoting the var values in the template, like:
- name: NEW_VAR_X
value: "${NEW_VAR_VALUE_X}"
No errors ever since.