General question: There's been quite a few PR's recently making istioctl behave like kubectl. Why not use kubectl directly?

@ayj My original inspiration for istioctl get improvements was so that there could be calculated custom fields summarizing deployment configuration. See #5538

I implemented the custom fields, but to make the PRs small and manageable I broke the implementation into several PRs and the last and most interesting one is not yet delivered.

I believe there is some easy low-hanging fruit with istioctl improvements. Today I spent 90 minutes debugging a clients 404 issues. Turns out an Istio VirtualService had been modified with kubectl edit bypassing Istio validation. After a bogus change was made kubectl get virtualservice showed no apparently problem, but istioctl get virtualservice showed

json: cannot unmarshal object into Go value of type []json.RawMessage

kubectl can't do any validation; at least with istioctl we now know something is wrong.

My hope is that a few short changes to istioctl will help medium-skilled operators from shooting themselves in the foot. (I realize the particular validation problem described here is better fixed by #1894 )

Thanks for taking time to improve the overall user experience. My comment was more in reference to issues (I mistakenly said PRs) that have suggested feature enhancements to bring istioctl in-line with kubectl, e.g.

• istioctl should support get multiple resources #6025 - istioctl should support get multiple resources
• Enable istioctl edit to edit some resources #6004 - Enable istioctl edit to edit some resources
• Introduce istioctl apply to update resources #5911 - Introduce istioctl apply to update resources

kubectl can't do any validation; at least with istioctl we now know something is wrong.

Server-side validation (#1894) will be enabled by default in 1.0 at which point we should remove client-side validation from istioctl. Server-side validation is also available in 0.8 but disabled by default.

Codecov Report

Merging #6013 into master will increase coverage by 1%.
The diff coverage is 64%.

@@          Coverage Diff           @@
##           master   #6013   +/-   ##
======================================
+ Coverage      67%     67%   +1%     
======================================
  Files         345     345           
  Lines       30684   30611   -73     
======================================
- Hits        20452   20444    -8     
+ Misses       9397    9330   -67     
- Partials      835     837    +2

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update dfd9364...1c44d75. Read the comment docs.

Closing; I will create a new PR after I find a fool-proof way to add a dependency.

Marking this WIP because I suspect the failure in ci/circleci: e2e-pilot-auth-v1alpha3-v2 is real. I introduce a new field, model.ConfigMeta.CreationTimestamp and I suspect that some code compares ConfigMeta with DeepEqual() and the timestamps cause the comparisons that previously returned true to become false.

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: ayj, esnible

The full list of commands accepted by this bot can be found here.

The pull request process is described here

@esnible: The following test failed, say /retest to rerun them all: