-
Notifications
You must be signed in to change notification settings - Fork 1.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Refactor the stack services
command to be uniform
#2131
Conversation
ea95cdd
to
f7b6ff3
Compare
Force pushed to add comments on exported functions |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, thank you @rumpl 👍
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I left some comments; one nit, and one (slightly) bigger change.
Code looks good in its current form, but (if it's not super-urgent to have this merged right away), I'd like to see if we can incorporate the changes from moby/moby#39231, so that we don't have to change the signature of these functions (twice)
// RunServices is the kubernetes implementation of docker stack services | ||
func RunServices(dockerCli *KubeCli, opts options.Services) error { | ||
// GetServices is the kubernetes implementation of listing stack services | ||
func GetServices(dockerCli *KubeCli, opts options.Services) ([]swarm.Service, map[string]service.ListInfo, error) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm looking at the extra return here; the service.ListInfo
is basically generated from information that's in []swarm.Service
, combined with information about the number of tasks running / desired. If I see it correctly, the service.ListInfo
here is needed, because swarm.Service
has no field for task status ("mode" is already part of the Service spec).
I just (finally) merged moby/moby#39231, which is a PR that adds exact those fields to the Service
struct, which means we could just return ([]swarmService, error)
, and have the presentation code either convert it to a ListInfo
or adjust that code to use []swarm.Service
directly.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let me prepare a vendor bump to bring in the changes from moby/moby#39231
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sure, I can wait for the vendor bump, thanks!
} | ||
stackName := opts.Namespace | ||
_, err = stacks.Get(stackName) | ||
if apierrs.IsNotFound(err) { | ||
return fmt.Errorf("nothing found in stack: %s", stackName) | ||
return []swarm.Service{}, nil, nil |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I was looking if we should return the error here, and have it handled by the caller 🤔. That way, this function (when used in different situations) would not hide this information.
I think that's ok for a follow-up, because we should have a function to convert Kubernetes API errors to errdef
errors. In that case we would map (if possible) / standardise these errors, so that the non-k8s specific code understands them;
cli/vendor/k8s.io/apimachinery/pkg/api/errors/errors.go
Lines 456 to 459 in 0b6685b
// IsNotFound returns true if the specified error was created by NewNotFound. | |
func IsNotFound(err error) bool { | |
return ReasonForError(err) == metav1.StatusReasonNotFound | |
} |
// statusCodeFromContainerdError returns status code for containerd errors when
// consumed directly (not through gRPC)
func statusCodeFromContainerdError(err error) int {
switch {
case containerderrors.IsInvalidArgument(err):
return http.StatusBadRequest
case containerderrors.IsNotFound(err):
return http.StatusNotFound
case containerderrors.IsAlreadyExists(err):
return http.StatusConflict
case containerderrors.IsFailedPrecondition(err):
return http.StatusPreconditionFailed
case containerderrors.IsUnavailable(err):
return http.StatusServiceUnavailable
case containerderrors.IsNotImplemented(err):
return http.StatusNotImplemented
default:
return http.StatusInternalServerError
}
}
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The reason I didn't return an error here is because the same code for swarm returns an empty slice when no services are found. We can do it one way or the other, we should just make sure that errors are returned in the same places for kubernetes and swarm.
f7b6ff3
to
ab837de
Compare
Rebased with master and changed the imports of |
Running `docker stack services <STACK> --orchestrator swarm would yield the message "Noting found in stack: asdf" with an exit code 0. The same command with kubernetes orchestrator would yield "nothing found in stack: adsf" (note the lower-case "nothing") and a non-zero exit code. This change makes the `stack services` command uniform for both orchestrators. The logic of getting and printing services is split to reuse the same formatting code. Signed-off-by: Djordje Lukic <djordje.lukic@docker.com>
ab837de
to
71eef01
Compare
carrying in #2167 - which is now ready for review 👍 |
👍 thanks a bunch! |
- What I did
Running
docker stack services <STACK> --orchestrator swarm
would yield the message "Noting found in stack: asdf" with an exit code 0. The same command with kubernetes orchestrator would yield "nothing found in stack: adsf" (note the lower-case "nothing") and a non-zero exit code. This change makes thestack services
command uniform for both orchestrators. The logic of getting and printing services is split to reuse the same formatting code.- How to verify it
Run:
Both commands should return
0
and show the error messageNoting found in stack: unknown
.- Description for the changelog
Stack services will return the same exit code for all orchestrators.
- A picture of a cute animal (not mandatory but encouraged)