add a new `--degraded` flag to `argocd app wait` command

[Cylix]

Summary

We would like to add a new flag to the argocd app wait CLI to stop waiting when the health status is degraded.

Motivation

argocd app wait already have the following flags:

• --operation: waits for pending operations.
• --sync: waits for the app to be synced
• --suspended: wait for the health status to be suspended, can be combined with --health
• --health: wait for the health status to be healthy, or to transitioned to degraded (from another known status).

For reference:

• https://github.com/argoproj/argo-cd/blob/master/cmd/argocd/commands/app.go#L1722
• https://github.com/argoproj/argo-cd/blob/5676ab5c9bc37cf4d8e60968745f30a3eeb98038/cmd/argocd/commands/app.go#L1606-L1620

We perform our deployments in our CI pipelines by running the following commands:

1. argocd app sync to trigger a sync
2. argo app wait --health --timeout <some timeout> to wait for the deployment to complete

This works for most cases, but there is one scenario where this does not work well. If the sync instantly fails, or fails very quickly (for example, canaries misconfiguration), argocd app wait will directly see the app in a degraded state and will wait until the timeout is reached, giving delayed feedback to our developers and generating a bit of confusion.

We tried to play with the currently available flags, but none is really fitting our need. The only workaround we found so far is to call argocd app get after the sync and before the wait: we can then grep the health status and decide whether or not to run argocd app wait. But overall, that'd be cleaner if we just have a combination of flags to exit argocd app wait if the app is either healthy or degraded.

Proposal

There is already a --suspended flag for argocd app wait to wait on the suspended health state. This flag can also be combined with --health to wait on both healthy or suspended state.

The proposal is to add a new flag --degraded to argocd app wait that works like --suspended:

• If using --degraded only, argocd app wait will wait until the state is degraded
• If using --health --degraded, argocd app wait will wait until the state is either degraded or healthy

[Cylix]

FYI, I opened a PR with the change I have in mind: https://github.com/argoproj/argo-cd/pull/8497

[lGSMl]

we faced exactly the same scenario in our CI/CD and we watch this issue with great interest. Is there a chance the PR will get a go in the following months? I see it is pending review, though if any help is required down the line with implementation I would be happy to assist.

[dhernandoatnewrelic]

I ran argocd app wait --help on v2.8.10 and the flag seems to be present. Shouldn't this issue be closed? See output:

Flags:
      --degraded               Wait for degraded
      --health                 Wait for health
  -h, --help                   help for wait
      --operation              Wait for pending operations
      --resource stringArray   Sync only specific resources as GROUP:KIND:NAME or !GROUP:KIND:NAME. Fields may be blank and '*' can be used. This option may be specified repeatedly
  -l, --selector string        Wait for apps by label. Supports '=', '==', '!=', in, notin, exists & not exists. Matching apps must satisfy all of the specified label constraints.
      --suspended              Wait for suspended
      --sync                   Wait for sync
      --timeout uint           Time out after this many seconds