CHT deploy script error: upgrade-service image referenced incorrectly

[1yuv]

Deploying with latest (master) cht-script results in incorrectly tagged pod and the service image is incorrect.

Steps to replicate:

1. Check out current master

2. Create values.yaml with this document reference

3. Apply this with ./cht-deploy script from cht-core/scripts/deploy.

4. When it completes applying, you get sucessful deployment.

5. Check the status of pods. upgrade-service is failed at InvalidImageName.

   ➜  deploy git:(master) ✗ k -n yuvraj-dev get po                                       
   NAME                                       READY   STATUS             RESTARTS   AGE
   cht-api-5f99b5684f-5kl44                   1/1     Running            0          9m11s
   cht-couchdb-ffd46779-hlq72                 1/1     Running            0          9m11s
   cht-haproxy-8f66cb488-5vzkx                1/1     Running            0          9m11s
   cht-haproxy-healthcheck-7fbc585b9c-bznrx   1/1     Running            0          9m11s
   cht-sentinel-56f958f5b4-62htm              1/1     Running            0          9m11s
   upgrade-service-8c894c85c-7jbvl            0/1     InvalidImageName   0          9m11s
   

6. When you inspect the pod, you get detailed error:

    state:
      waiting:
        message: 'Failed to apply default image tag "medicmobile/upgrade-service:medicmobile/upgrade-service:0.32":
          couldn''t parse image reference "medicmobile/upgrade-service:medicmobile/upgrade-service:0.32":
          invalid reference format'
        reason: InvalidImageName
   

7. The reason for this is image is being referenced incorectly.

   spec:
     containers:
     - env:
       - name: CHT_NAMESPACE
         valueFrom:
           fieldRef:
             apiVersion: v1
             fieldPath: metadata.namespace
       - name: CHT_DEPLOYMENT_NAME
         valueFrom:
           fieldRef:
             apiVersion: v1
             fieldPath: metadata.labels['cht.service']
       - name: UPGRADE_SERVICE_PORT
         value: "5008"
       image: medicmobile/upgrade-service:medicmobile/upgrade-service:0.32
       imagePullPolicy: IfNotPresent
       name: upgrade-service
   

[1yuv]

This could possibly be a bug in medic/helm-charts.

The correct image tag for this deployment's container definition should be:

image: medicmobile/upgrade-service:medicmobile/upgrade-service:0.32

I was able to run upgrade-service after I edited deployment and pointed to correct image tag.

[henokgetachew]

This PR fixes it. It will be released today.

[dianabarsan]

@1yuv Can you please verify that this has been resolved?

[1yuv]

This is not fixed yet. I ran helm repo update and applied the changes just as specified here, and the issue is same.

NAME                                       READY   STATUS             RESTARTS   AGE
cht-api-5f99b5684f-4vwx4                   1/1     Running            0          11s
cht-couchdb-ffd46779-9n4z6                 1/1     Running            0          11s
cht-haproxy-8f66cb488-c6nt8                1/1     Running            0          11s
cht-haproxy-healthcheck-7fbc585b9c-p2869   1/1     Running            0          10s
cht-sentinel-56f958f5b4-9b7ts              1/1     Running            0          10s
upgrade-service-8c894c85c-4d4w8            0/1     InvalidImageName   0          11s

[latin-panda]

I had to start an instance to check the CHT release quickly and experienced this exact issue today

[henokgetachew]

The fix is in the cht-deploy-fixes branch @latin-panda. Can you confirm that works for you?

[Benmuiruri]

@henokgetachew I can confirm switching to that branch and deploying from there worked.

[henokgetachew]

FYI you can also npx to deploy now. It's been published to npm. Meaning you don't have to do it from the repo. Just run npx @medic/cht-deploy -f <path-from-your-yaml>

[henokgetachew]

I will leave the ticket open until the PR gets merged. I'll leave it assigned to myself until then.

[1yuv]

Hi @henokgetachew , associated PR appears to be merged. Based on this, can you close this issue?