feat: introduce v1beta1 types #1073

mehabhalodiya · 2022-03-26T05:40:34Z

Signed-off-by: mehabhalodiya mehabhalodiya@gmail.com

What type of PR is this?
/kind feature

What this PR does / why we need it:
As suggested by @robscott, this is a follow-up PR on one of our last steps before the v1beta1 (v0.5.0) release.
The key is that we’d need GatewayClass, Gateway, and HTTPRoute type definitions in apis/v1beta1 in addition to the alpha directories they already exist in.

Which issue(s) this PR fixes:
Fixes #1041

Does this PR introduce a user-facing change?:

Introduce v1beta1 types for resources graduating to beta

Signed-off-by: mehabhalodiya <mehabhalodiya@gmail.com>

k8s-ci-robot · 2022-03-26T05:40:48Z

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: mehabhalodiya
To complete the pull request process, please assign robscott after the PR has been reviewed.
You can assign the PR to them by writing /assign @robscott in a comment when ready.

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

Needs approval from an approver in each of these files:

OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

mehabhalodiya · 2022-03-26T05:59:03Z

/cc @robscott

I fooled myself into thinking this wasn't actually doing anything, so I figure it would be good to add a reassuring success message: $ ./hack/verify-boilerplate.sh Verified 111 file headers match boilerplate

Add a success message to verify-boilerplate

akutz

Thank you for submitting this PR! I am not finished with my review, but I have to go for now. I will come back and finish looking through the PR as soon as I am able. Certainly by the end of tomorrow.

akutz · 2022-03-30T22:17:53Z

apis/v1beta1/gateway_types.go

+type GatewaySpec struct {
+	// GatewayClassName used for this Gateway. This is the name of a
+	// GatewayClass resource.
+	GatewayClassName ObjectName `json:"gatewayClassName"`


Is there a reason to not just name the field ClassName? Or even just GatewayClass? The latter would be consistent with Service.Spec.LoadBalancerClass. That said, the Name suffix is used by Ingress.Spec.IngressClassName. In fact, the Name suffix is also used by PersistentVolumeClaim.Spec.StorageClassName, so it appears as if the outlier is Service.Spec.LoadBalancerClass and the convention is to use the Name suffix. Carry on 😃

Also, the comment could be simplified:

// GatewayClassName is the name of the GatewayClass resource used by // this Gateway.

Finally, the use of ObjectName is interesting. I see it has a validation on it for min and max length. However, the comments on the type also indicate additional, possible validations:

// ObjectName refers to the name of a Kubernetes object. // Object names can have a variety of forms, including RFC1123 subdomains, // RFC 1123 labels, or RFC 1035 labels. // // +kubebuilder:validation:MinLength=1 // +kubebuilder:validation:MaxLength=253

I wonder if there is a reason to not just use string and let the webhook that will likely do the additional validation also be responsible for the min/max length validation?

Hey @akutz! We're coming to the end of the v1beta1 release window, and both @robscott and I realized that your feedback had slipped through the cracks. So, I'll respond now where I think we probably won't make changes, and make issues for ones where we will.

For the min-max length validations and others, the general guideline we're using is to only add custom validation when we can't get it from the kubebuilder/OpenAPI validation.

akutz · 2022-03-30T22:26:55Z

apis/v1beta1/gateway_types.go

+	//
+	// Support: Core
+	//
+	// +listType=map


Ah, cool. I have not seen this used yet, so it's neat to see it!

akutz · 2022-03-30T22:27:14Z

apis/v1beta1/gateway_types.go

+	// +listType=map
+	// +listMapKey=name
+	// +kubebuilder:validation:MinItems=1
+	// +kubebuilder:validation:MaxItems=64


So this is odd to me. Why a max of 64 listeners? That seems arbitrary?

We haven't written this down anywhere, I don't think, but we prefer not to have unbounded lists, since people will inevitably try to stuff 1000 things in there.

For times when we don't have a good idea, we do just pick an arbitrary limit that seems larger than what we think most uses will need, and accept that we may need to increase later. This is the case for all the limits you've called out here. Yes, they are arbitrary, but we needed a value >1 and < infinity, and we have no data about actual usage, so any number we pick is arbitrary.

I'll summarize this in other replies as "No unbounded lists".

I just realized that I didn't make clear that the other reason for the limits is the max record size in etcd, which defaults to about 1MB of text in deserialized form.

akutz · 2022-03-30T22:29:00Z

apis/v1beta1/gateway_types.go

+	//
+	// +listType=map
+	// +listMapKey=name
+	// +kubebuilder:validation:MinItems=1


I do not think a non-zero list should be required to set up a gateway. I can easily see a gateway being created to reserve or grab an IP address before any listeners are defined. I would like there to be an allowed min items of zero and mark this field as +optional and omitempty. Maybe there was some discussion about this at one point that I missed? Thanks!

akutz · 2022-03-30T22:30:10Z

apis/v1beta1/gateway_types.go

+	// Support: Core
+	//
+	// +optional
+	// +kubebuilder:validation:MaxItems=16


Once again the use of a max here seems arbitrary. Why would this be limited?

See "No unbounded lists" comment.

akutz · 2022-03-30T22:49:02Z

apis/v1beta1/gateway_types.go

+	// Group is the group of the Route.
+	//
+	// +optional
+	// +kubebuilder:default=gateway.networking.k8s.io


In any other case an empty group would default to v1. I get why it is defaulting to the group you have, but should it?

The group is the domain part of the apiVersion field, v1 is the version. We've chosen to leave out version here to make it so that users don't have to manually update versions as we move through them - on the assumption we'll move through versions much faster than upstream Kubernetes does.

akutz · 2022-03-30T22:50:12Z

apis/v1beta1/gateway_types.go

+}
+
+// RouteGroupKind indicates the group and kind of a Route resource.
+type RouteGroupKind struct {


I think this should be renamed to RouteGroupVersionKind and enforce the use of the version in this struct as well. Types at specific versions can sometimes be entirely different types, and we need to know the version of the type we're selecting.

akutz · 2022-03-30T22:51:07Z

apis/v1beta1/gateway_types.go

+	// Type of the address.
+	//
+	// +optional
+	// +kubebuilder:validation:Enum=IPAddress;Hostname;NamedAddress


Why is the validation not on the type AddressType string declaration just like it is for type FromNamespaces string and other declarations in this PR?

Also, I think we may want to address dual-stack and explicit IPv6 before going beta. This has been a pain in Kube proper for a while, and for the GW APIs to not have parity would be a step back IMO.

I've addressed this and other concerns in #1178.

akutz · 2022-03-30T22:53:50Z

apis/v1beta1/gateway_types.go

+	// assigns an address from a reserved pool.
+	//
+	// +optional
+	// +kubebuilder:validation:MaxItems=16


Again, why this seemingly arbitrary max?

See "No unbounded lists" comment.

akutz · 2022-03-30T22:54:15Z

apis/v1beta1/gateway_types.go

+	// +optional
+	// +listType=map
+	// +listMapKey=name
+	// +kubebuilder:validation:MaxItems=64


Again, why this seemingly arbitrary max?

See "No unbounded lists" comment.

…> Gateways attachment.

hbagdi · 2022-04-01T22:49:27Z

It may be best to not review such humungous PRs and find other alternatives to review the patch-diff between v1alpha2 and v1beta1. Thoughts?

robscott · 2022-04-01T22:54:01Z

Yep, sorry for the confusion on this PR! These are great comments @akutz and I'll try to respond to them soon, but this particular PR was actually intended to be more of a checklist item on the milestone than the actual API review. I've been super busy, but my goal is to get a PR out that is just a diff of aps/v1alpha2 between v0.4.0 release and now, like @hbagdi suggested above. Of course if anyone else wants to take that on, I'm happy to let someone else own it. Either way, hopefully we can have a smaller diff ready for review on Monday.

akutz · 2022-04-02T00:56:48Z

It may be best to not review such humungous PRs

@hbagdi, I absolutely know what you mean, but at the same time I could not help but smile at just this part 😄

mehabhalodiya · 2022-04-02T03:29:30Z

@hbagdi @robscott I have not done it correctly. Right?
I will make changes according to your comments. Sorry for this!
I am a newbie and trying to understand, it would be great if you people help me 🙂

robscott · 2022-04-02T04:38:41Z

Hey @mehabhalodiya, sorry for the confusion! Your PR is headed in the right direction, I'd just missed creating the corresponding PR that compared v0.4.0 and master API types for the sake of a simpler API review. I'm hoping to have that in place on Monday, but will also work to respond to the great feedback here as well.

akutz · 2022-04-02T20:23:36Z

If y'all want to ignore my feedback on this PR, I will post it on the newer PR when Robb posts it. I will still wait to resolve this PR's feedback until the other is posted, just so it is a little easier for me to find the comments.

Provide more details on route parentRef in documentation for routes <> Gateways attachment.

…-core-tls-support clarify core support of secret reference

Bumping version references to v0.4.3

Adds the status of the "Accepted" condition to the GatewayClass print columns list. Signed-off-by: Steve Kriss <krisss@vmware.com>

- fixes path prefix match YAML - removes check for HTTPRoute to be Accepted pending kubernetes-sigs#1112 Signed-off-by: Steve Kriss <krisss@vmware.com>

Signed-off-by: Steve Kriss <krisss@vmware.com>

…policy-fix conformance: fixes for invalid reference policy test

fix: typo for ambassador API gateway

…pted-printcolumn add GatewayClass Accepted print column

Cleaning up godocs after v0.5.0 API Review

The GatewayTLSConfig CertificateRefs member is a slice of pointer to struct which implies that the struct is optional. Change this to a slice of struct to better fit with API type conventions.

We would like our example YAML files to include comments that do not show up in the guides where the files are used as jinja templates, while still ensuring they are valid YAML files. Configuring the jinja Environment with "line_comment_prefix" allows you to control the syntax of the jinja comment directive. If we set that to "##", then such lines are ommitted during jinja templating, and are treated as YAML comments. See also https://jinja.palletsprojects.com/en/2.11.x/templates/#line-statements

Add Jinja "line comments" to our example manifests which reference the guides which use them. Generated using this awfulness: $ sed -i '1s/^/## Used in:\n/' $( \ for iii in $(git grep -l '{% include' site-src/); do \ awk -F "'" '/{% include/ { print "examples/"$2 }' $iii ; \ done | sort | uniq) $ for iii in $(git grep -l '{% include' site-src/); do \ for jjj in $(awk -F "'" '/{% include/ { print "examples/"$2 }' $iii); do \ sed -i "2s|^|## - $iii\n|" $jjj; \ done; \ done Co-authored-by: Mark McLoughlin <markmc@redhat.com>

…s-backlinks docs: backlinks from example manifests to guides

Fix certificateRefs optionality.

This doc's intended audience is users rather than implementors, so these keywords aren't really appropriate.

docs: add note on RFC2119 keyword semantics

…me-spec Clarifying that wildcard hostname matching is suffix matching

…odifier Adding validation to ensure prefix path match exists for path modifier

…e-policy Renaming ReferencePolicy to ReferenceGrant

Signed-off-by: mehabhalodiya <mehabhalodiya@gmail.com>

mehabhalodiya · 2022-06-07T07:12:48Z

Sorry for the delay!
I am re-copying the files from v1alpha2, just give me a moment!
Further, Nick and Rob were going to split up responding to the comments from Andrew.
Thank you for your patience 🙂

youngnick · 2022-06-07T07:25:38Z

I think I've hit most of Andrew's concerns.

I think that we should actually get this merged into master asap, and then if we need to make any further v1alpha2 changes, we copy them over.

Signed-off-by: mehabhalodiya <mehabhalodiya@gmail.com>

feat: introduce v1beta1 types

205398a

Signed-off-by: mehabhalodiya <mehabhalodiya@gmail.com>

k8s-ci-robot added kind/feature Categorizes issue or PR as related to a new feature. cncf-cla: yes Indicates the PR's author has signed the CNCF CLA. size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files. labels Mar 26, 2022

k8s-ci-robot requested review from hbagdi and jpeach March 26, 2022 05:40

k8s-ci-robot requested a review from robscott March 26, 2022 05:59

markmc and others added 6 commits March 29, 2022 18:29

Add a success message to verify-boilerplate

04cdf5d

I fooled myself into thinking this wasn't actually doing anything, so I figure it would be good to add a reassuring success message: $ ./hack/verify-boilerplate.sh Verified 111 file headers match boilerplate

Support eventual consistency in request-response helper

2fb0f78

Use request-response helper for all tests

85d58a1

Log expected vs. actual while waiting for consistency

824539b

Assert consistency once initial request succeeds

d01bd47

Merge pull request kubernetes-sigs#1079 from markmc/boilerplate-success

bae59a6

Add a success message to verify-boilerplate

akutz suggested changes Mar 30, 2022

View reviewed changes

nathancoleman and others added 3 commits April 1, 2022 15:12

Allow for intermittent success prior to reaching consistency

41097d5

Log when we've reached the consistency threshold for a request

736fd46

Provide more details on route parentRef in documentation for routes <…

5dbaa40

…> Gateways attachment.

jakubdyszkiewicz and others added 5 commits April 4, 2022 10:13

clarify core support of secret reference

cd12a93

Add NGINX Kubernetes Gateway to list of implementations

f56d68b

Merge pull request kubernetes-sigs#1082 from cxhiano/1021-doc

3ec7311

Provide more details on route parentRef in documentation for routes <> Gateways attachment.

fix autogenerated files

0c30bed

Merge pull request kubernetes-sigs#1084 from jakubdyszkiewicz/clarify…

0a4a543

…-core-tls-support clarify core support of secret reference

k8s-ci-robot and others added 25 commits May 19, 2022 02:30

Merge pull request kubernetes-sigs#1150 from robscott/v0.4.3-bump

c7fe7e4

Bumping version references to v0.4.3

add GatewayClass Accepted print column

1e53239

Adds the status of the "Accepted" condition to the GatewayClass print columns list. Signed-off-by: Steve Kriss <krisss@vmware.com>

fix: typo for ambassador API gateway

95cb5ff

conformance: fixes for invalid reference policy test

903eeb2

- fixes path prefix match YAML - removes check for HTTPRoute to be Accepted pending kubernetes-sigs#1112 Signed-off-by: Steve Kriss <krisss@vmware.com>

move & update TODO comment

da17817

Signed-off-by: Steve Kriss <krisss@vmware.com>

Merge pull request kubernetes-sigs#1170 from skriss/conf-invalid-ref-…

9b2c137

…policy-fix conformance: fixes for invalid reference policy test

Merge pull request kubernetes-sigs#1169 from gabel/patch-1

34cb4c7

fix: typo for ambassador API gateway

Merge pull request kubernetes-sigs#1168 from skriss/gatewayclass-acce…

50e6186

…pted-printcolumn add GatewayClass Accepted print column

Merge pull request kubernetes-sigs#1160 from robscott/godoc-fixes

66a7c8f

Cleaning up godocs after v0.5.0 API Review

Fix certificateRefs optionality.

76eb71e

The GatewayTLSConfig CertificateRefs member is a slice of pointer to struct which implies that the struct is optional. Change this to a slice of struct to better fit with API type conventions.

Merge pull request kubernetes-sigs#1177 from markmc/examples-to-guide…

d48f7c4

…s-backlinks docs: backlinks from example manifests to guides

Merge pull request kubernetes-sigs#1176 from briantkennedy/pointerslice

5ca7170

Fix certificateRefs optionality.

docs: add note on RFC2119 keyword semantics

20e2621

docs: don't use RFC2119 keywords in API overview

74617a9

This doc's intended audience is users rather than implementors, so these keywords aren't really appropriate.

Merge pull request kubernetes-sigs#1174 from markmc/rfc2119

d1b8664

docs: add note on RFC2119 keyword semantics

Adding validation to ensure prefix path match exists for path modifier

65271b1

Clarifying that wildcard hostname matching is suffix matching

2672b1a

Merge pull request kubernetes-sigs#1173 from robscott/wildcard-hostna…

c5bca72

…me-spec Clarifying that wildcard hostname matching is suffix matching

Merge pull request kubernetes-sigs#1171 from robscott/validate-path-m…

20683af

…odifier Adding validation to ensure prefix path match exists for path modifier

Renaming ReferencePolicy to ReferenceGrant

d1370a7

Merge pull request kubernetes-sigs#1179 from robscott/rename-referenc…

f3bb40f

…e-policy Renaming ReferencePolicy to ReferenceGrant

feat: introduce v1beta1 types

fd06335

Signed-off-by: mehabhalodiya <mehabhalodiya@gmail.com>

recopy the types

f6b5b15

Signed-off-by: mehabhalodiya <mehabhalodiya@gmail.com>

resolved merge conflicts

196dde1

Signed-off-by: mehabhalodiya <mehabhalodiya@gmail.com>

mehabhalodiya closed this Jun 7, 2022

mehabhalodiya deleted the v1beta1-types branch June 7, 2022 17:41

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: introduce v1beta1 types #1073

feat: introduce v1beta1 types #1073

mehabhalodiya commented Mar 26, 2022

k8s-ci-robot commented Mar 26, 2022

mehabhalodiya commented Mar 26, 2022

akutz left a comment

akutz Mar 30, 2022

youngnick Jun 7, 2022

akutz Mar 30, 2022

akutz Mar 30, 2022

youngnick Jun 7, 2022 •

edited

Loading

youngnick Jun 7, 2022

akutz Mar 30, 2022

akutz Mar 30, 2022

youngnick Jun 7, 2022

akutz Mar 30, 2022

youngnick Jun 7, 2022

akutz Mar 30, 2022

akutz Mar 30, 2022

youngnick Jun 7, 2022

akutz Mar 30, 2022

youngnick Jun 7, 2022

akutz Mar 30, 2022

youngnick Jun 7, 2022

hbagdi commented Apr 1, 2022

robscott commented Apr 1, 2022

akutz commented Apr 2, 2022

mehabhalodiya commented Apr 2, 2022

robscott commented Apr 2, 2022

akutz commented Apr 2, 2022

mehabhalodiya commented Jun 7, 2022

youngnick commented Jun 7, 2022

feat: introduce v1beta1 types #1073

feat: introduce v1beta1 types #1073

Conversation

mehabhalodiya commented Mar 26, 2022

k8s-ci-robot commented Mar 26, 2022

mehabhalodiya commented Mar 26, 2022

akutz left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

youngnick Jun 7, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hbagdi commented Apr 1, 2022

robscott commented Apr 1, 2022

akutz commented Apr 2, 2022

mehabhalodiya commented Apr 2, 2022

robscott commented Apr 2, 2022

akutz commented Apr 2, 2022

mehabhalodiya commented Jun 7, 2022

youngnick commented Jun 7, 2022

youngnick Jun 7, 2022 •

edited

Loading