Managing State with Deployments kubectl --record unnown flag

GrzegorzArtichowicz

Hi,

I have problem with flag --record ? I don't know whats the point.
kubectl create deploy ghost --image=ghost --record
Error: unknown flag: --record

kubectl version
Client Version: version.Info{Major:"1", Minor:"19", GitVersion:"v1.19.4",

chrispokorni

Hi @GrzegorzArtichowicz,

You can try replacing create with apply.

Regards ,
-Chris

GrzegorzArtichowicz

Hi @chrispokorni

Yes I know but I would like to know why doesn't working ,excercise is in "imperative" way
Best,
GA

chrispokorni

Hi @GrzegorzArtichowicz,

Features are often changing in Kubernetes as the software matures and it goes through several releases. You can check the documentation or the command line reference to see whether there is any explanation why the flag no longer works in the given context. If there are no references to it being deprecated or no longer supported (and it is expected to work), then this unexpected behavior could be reported as a bug on the Kubernetes github issues site.

Regards,
-Chris

GrzegorzArtichowicz

ok ,thank you

chaosreload

@GrzegorzArtichowicz said:
Hi,

I have problem with flag --record ? I don't know whats the point.
kubectl create deploy ghost --image=ghost --record
Error: unknown flag: --record

kubectl version
Client Version: version.Info{Major:"1", Minor:"19", GitVersion:"v1.19.4",

I got the same error with the client version v1.21.4: 'Client Version: version.Info{Major:"1", Minor:"21", GitVersion:"v1.21.4"'
there is no description of the error in 'kubectl create --help', but I found the 2 issues in the repo mentioned:
https://github.com/kubernetes/kubernetes/issues/81289
https://github.com/kubernetes/kubernetes/issues/40422
They show me that --record will be deprecated.
However, it is still in progress and should not affect in v1.21.4, I think this might be a bug?

chaosreload

@GrzegorzArtichowicz said:
Hi,

I have problem with flag --record ? I don't know whats the point.
kubectl create deploy ghost --image=ghost --record
Error: unknown flag: --record

kubectl version
Client Version: version.Info{Major:"1", Minor:"19", GitVersion:"v1.19.4",

I got the same error with the client version v1.21.4: 'Client Version: version.Info{Major:"1", Minor:"21", GitVersion:"v1.21.4"'
there is no description of the error in 'kubectl create --help', but I found the 2 issues in the repo mentioned:
https://github.com/kubernetes/kubernetes/issues/81289
https://github.com/kubernetes/kubernetes/issues/40422
They show me that --record will be deprecated.
However, it is still in progress and should not affect in v1.21.4, I think this might be a bug?

mgerstaedt

Hi there,
I've got this issue too. Could you please review Chapter 7 "Deployment Rollbacks"? It seems to be deprecated until 1.6, but the --record flag it still in the documentation. Therefore it's very confusing to me and maybe others.

Edit:
Also the CHANGE-CAUSE seems to not recorded anymore.

Thanks a lot.

chrispokorni

Hi @mgerstaedt,

Have you tried the --record=true syntax instead? It helps to record the change-cause annotation for a rolling update.

Regards,
-Chris

oleksazhel

I suggest to change this error in training materials, because --record flag is not available since 2020 at least.
As a workaround you can:

• Run kubectl create deploy ghost --image=ghost
• Run kubectl get deployments.apps ghost -o yaml > ghost.yaml
• Run vim ghost.yaml and remove status, time, versions related info
• Run kubectl delete deployments.apps ghost
• And now you can use flag record with warning that it's deprecated and will be removed in the future kubectl create -f ghost.yaml --record

pnts

I'm using v1.23.1 because that's what will be used in the CKA exam.

I do not get kubectl create --record to work.

However, if I do like this:
1. kubectl create -f my_deployment.yaml --save-config
2. Then make some change to my_deployment.yaml, like changing replica count.
3. kubectl apply -f my_deployment.yaml --record

It works but I get the deprecated warning:

Flag --record has been deprecated, --record will be removed in the future

kubectl set image is working with the --record flag.

I follow the discussion on Kubernetes Github where deprecation of --record flag is being discussed.
Manual annotation seems like one way of moving forward:
https://github.com/kubernetes/kubernetes/issues/40422#issuecomment-1118042103

oleksazhel

@pnts I believe you can do it in one step, just kubectl apply -f my_deployment.yaml --record or imperatively kubectl apply -f my_deployment.yaml --record or even with migration to declarative form kubectl apply -f my_deployment.yaml --record --save-settings. By the way why do you think that CKA exam will use 1.23.1 and not 1.23.8 for example?)

pnts

@oleksazhel
Yes, to do it declaratively worked. Like so:
kubectl apply -f my_deployment.yaml --record
Thanks!

As for kubernetes version, I got it wrong.
I was referring to this page:
https://training.linuxfoundation.org/certification/certified-kubernetes-administrator-cka/

But it says:

Software Version: Kubernetes v1.23

So I guess it could be any patch version.

jsm3031

Please take this as constructive criticism. I think it is unacceptable that for paid training, years later, it is not updated. I work full time and have kids and I don't have time to troubleshoot things that should be updated. I don't pay for courses to learn how to trouble shoot, i do that all day in my job. I need my study time to be very efficient. I get up at 5 am or earlier to have a bit of time to study and i don't want to waste time troubleshooting incorrect steps in the material. Please update the material

jsm3031

PS i think this is the best one-liner replacement for the command in the material

kubectl create deployment --image=ghost ghost --dry-run=client -o yaml | k create --record=true -f -

I have to admit I learned more about kubectl figuring that out.

michael.weinrich

Also ghost doesn't start for me as it obviously looks for a MySQL database on localhost...this is really bad for a paid course. This isn't what I expected.

ffol

Hi all,

I myself went into the same situation where the instructions are still not updated, however that should not trigger any extensive troubleshooting as the idea is to experience how the rollout works on a specific deployments. You may use the image in the previous exercise, something like this one:

kubectl create deploy dev-web --image=nginx:1.13.7-alpine --dry-run=client -o yaml | kubectl create --record=true -f -

kubectl set image deployment/dev-web nginx=nginx:1.13.8-alpine --all

kubectl rollout history deployment/dev-web

kubectl rollout undo deployment/dev-web ; kubectl get pods

kubectl rollout pause deployment/dev-web

kubectl rollout resume deployment/dev-web

Happy k8s learning!
Florin

robertodmaggi

Com'on man!
it's 2023, we paid so much money and the errors are still here?!
has anybody read those forums in three years?
I don't have time to waste to troubleshoot those so expensive courses!
I arrived here after one hour googleing, I study in the night, having full day work and two children.