Skip to content

Commit

Permalink
config/v1/types_cluster_version: Add conditionalUpdates and friends
Browse files Browse the repository at this point in the history 
Implementing targeted edge blocking as described in [1].  Except:

* Some properties which are optional in the enhancement are required
  here.  I think this is strictly worse handling for the
  cluster-version operator, because we are just getting this data from
  the upstream update service's Cincinnati JSON, and if the upstream
  says "here's a risk, but I am not linking a URI", and the CVO
  subsequently rejects the entire update payload as invalid, then
  downstream consumers are not hearing about the
  reason/message/matchingRules that the upstream may have been passing
  along.  But I'm not an API approver, and David is asking for the
  +required in [2,3,4,5].  I dunno why he isn't finding my argument's
  convincing, but I'm out of time to continue arguing my position.

* Similarly, the ClusterCondition.Type property has an enum, because
  David requires it [6].  This means that the CVO will need to
  self-censor if there are unrecognized types (because we don't want
  our ClusterVersion writes rejected on API-server validation
  failures).  That means that instead of saying "... and there was
  another rule $UNRECOGNIZED_TYPE that I don't know how to evaluate",
  the CVO will have to silently drop that rule.  And if that's the
  only rule, the CVO will have to silently drop that risk and
  conditional update.  David is fine with that, even if I'm not on
  board, and he's the API approver, so that's what we'll do.

  I've also upcased the types to Always and PromQL; from David:

    Always and PromQL.  We use CamelCase for enumerated values

* history[].overrides became history[].acceptedRisks to satisfy
  David's concerns about misinterpretation [7].

Swagger, deepcopy, and CRD changes were generated with:

  $ make update

[1]: openshift/enhancements#821
[2]: openshift#1011 (comment)
[3]: openshift#1011 (comment)
[4]: openshift#1011 (comment)
[5]: openshift#1011 (comment)
[6]: openshift#1011 (comment)
[7]: openshift#1011 (comment)
  • Loading branch information
@wking
wking committed Sep 23, 2021
1 parent 3cd8091 commit 03d33da
Show file tree
Hide file tree
Showing 4 changed files with 406 additions and 6 deletions.
134 changes: 133 additions & 1 deletion config/v1/0000_00_cluster-version-operator_01_clusterversion.crd.yaml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -112,7 +112,7 @@ spec:
- versionHash
properties:
availableUpdates:
description: availableUpdates contains the list of updates that are appropriate for this cluster. This list may be empty if no updates are recommended, if the update service is unavailable, or if an invalid channel has been specified.
description: availableUpdates contains updates recommended for this cluster. Updates which appear in conditionalUpdates but not in availableUpdates may expose this cluster to known issues. This list may be empty if no updates are recommended, if the update service is unavailable, or if an invalid channel has been specified.
type: array
items:
description: Release represents an OpenShift release image and associated metadata.
Expand All @@ -133,6 +133,135 @@ spec:
description: version is a semantic versioning identifying the update version. When this field is part of spec, version is optional if image is specified.
type: string
nullable: true
conditionalUpdates:
description: conditionalUpdates contains the list of updates that may be recommended for this cluster if it meets specific required conditions. Consumers interested in the set of updates that are actually recommended for this cluster should use availableUpdates. This list may be empty if no updates are recommended, if the update service is unavailable, or if an empty or invalid channel has been specified.
type: array
items:
description: ConditionalUpdate represents an update which is recommended to some clusters on the version the current cluster is reconciling, but which may not be recommended for the current cluster.
type: object
required:
- release
- risks
properties:
conditions:
description: 'conditions represents the observations of the conditional update''s current status. Known types are: * Evaluating, for whether the cluster-version operator will attempt to evaluate any risks[].matchingRules. * Recommended, for whether the update is recommended for the current cluster.'
type: array
items:
description: "Condition contains details for one aspect of the current state of this API Resource. --- This struct is intended for direct use as an array at the field path .status.conditions. For example, type FooStatus struct{ // Represents the observations of a foo's current state. // Known .status.conditions.type are: \"Available\", \"Progressing\", and \"Degraded\" // +patchMergeKey=type // +patchStrategy=merge // +listType=map // +listMapKey=type Conditions []metav1.Condition `json:\"conditions,omitempty\" patchStrategy:\"merge\" patchMergeKey:\"type\" protobuf:\"bytes,1,rep,name=conditions\"` \n // other fields }"
type: object
required:
- lastTransitionTime
- message
- reason
- status
- type
properties:
lastTransitionTime:
description: lastTransitionTime is the last time the condition transitioned from one status to another. This should be when the underlying condition changed. If that is not known, then using the time when the API field changed is acceptable.
type: string
format: date-time
message:
description: message is a human readable message indicating details about the transition. This may be an empty string.
type: string
maxLength: 32768
observedGeneration:
description: observedGeneration represents the .metadata.generation that the condition was set based upon. For instance, if .metadata.generation is currently 12, but the .status.conditions[x].observedGeneration is 9, the condition is out of date with respect to the current state of the instance.
type: integer
format: int64
minimum: 0
reason:
description: reason contains a programmatic identifier indicating the reason for the condition's last transition. Producers of specific condition types may define expected values and meanings for this field, and whether the values are considered a guaranteed API. The value should be a CamelCase string. This field may not be empty.
type: string
maxLength: 1024
minLength: 1
pattern: ^[A-Za-z]([A-Za-z0-9_,:]*[A-Za-z0-9_])?$
status:
description: status of the condition, one of True, False, Unknown.
type: string
enum:
- "True"
- "False"
- Unknown
type:
description: type of condition in CamelCase or in foo.example.com/CamelCase. --- Many .condition.type values are consistent across resources like Available, but because arbitrary conditions can be useful (see .node.status.conditions), the ability to deconflict is important. The regex it matches is (dns1123SubdomainFmt/)?(qualifiedNameFmt)
type: string
maxLength: 316
pattern: ^([a-z0-9]([-a-z0-9]*[a-z0-9])?(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*/)?(([A-Za-z0-9][-A-Za-z0-9_.]*)?[A-Za-z0-9])$
x-kubernetes-list-map-keys:
- type
x-kubernetes-list-type: map
release:
description: release is the target of the update.
type: object
properties:
channels:
description: channels is the set of Cincinnati channels to which the release currently belongs.
type: array
items:
type: string
image:
description: image is a container image location that contains the update. When this field is part of spec, image is optional if version is specified and the availableUpdates field contains a matching version.
type: string
url:
description: url contains information about this release. This URL is set by the 'url' metadata property on a release or the metadata returned by the update API and should be displayed as a link in user interfaces. The URL field may not be set for test or nightly releases.
type: string
version:
description: version is a semantic versioning identifying the update version. When this field is part of spec, version is optional if image is specified.
type: string
risks:
description: risks represents the range of issues associated with updating to the target release. The cluster-version operator will evaluate all entries, and only recommend the update if there is at least one entry and all entries recommend the update.
type: array
minItems: 1
items:
description: ConditionalUpdateRisk represents a reason and cluster-state for not recommending a conditional update.
type: object
required:
- matchingRules
- message
- name
- url
properties:
matchingRules:
description: matchingRules is a slice of conditions for deciding which clusters match the risk and which do not. The slice is ordered by decreasing precedence. The cluster-version operator will walk the slice in order, and stop after the first it can successfully evaluate. If no condition can be successfully evaluated, the update will not be recommended.
type: array
minItems: 1
items:
description: ClusterCondition is a union of typed cluster conditions. The 'type' property determines which of the type-specific properties are relevant. When evaluated on a cluster, the condition may match, not match, or fail to evaluate.
type: object
required:
- type
properties:
promql:
description: promQL represents a cluster condition based on PromQL.
type: object
required:
- promql
properties:
promql:
description: PromQL is a PromQL query classifying clusters. This query query should return a 1 in the match case and a 0 in the does-not-match case case. Queries which return no time series, or which return values besides 0 or 1, are evaluation failures.
type: string
type:
description: type represents the cluster-condition type. This defines the members and semantics of any additional properties.
type: string
enum:
- Always
- PromQL
message:
description: message provides additional information about the risk of updating, in the event that matchingRules match the cluster state. This is only to be consumed by humans. It may contain Line Feed characters (U+000A), which should be rendered as new lines.
type: string
minLength: 1
name:
description: name is the CamelCase reason for not recommending a conditional update, in the event that matchingRules match the cluster state.
type: string
minLength: 1
url:
description: url contains information about this risk.
type: string
format: uri
minLength: 1
x-kubernetes-list-map-keys:
- name
x-kubernetes-list-type: map
conditions:
description: conditions provides information about the cluster version. The condition "Available" is set to true if the desiredUpdate has been reached. The condition "Progressing" is set to true if an update is being applied. The condition "Degraded" is set to true if an update is currently blocked by a temporary or permanent error. Conditions are only valid for the current desiredUpdate when metadata.generation is equal to status.generation.
type: array
Expand Down Expand Up @@ -191,6 +320,9 @@ spec:
- state
- verified
properties:
acceptedRisks:
description: acceptedRisks records risks which were accepted to initiate the update. For example, it may menition an Upgradeable=False or missing signature that was overriden via desiredUpdate.force, or an update that was initiated despite not being in the availableUpdates set of recommended update targets.
type: string
completionTime:
description: completionTime, if set, is when the update was fully applied. The update that is currently being applied will have a null completion time. Completion time will always be set for entries that are not the current update (usually to the started time of the next update).
type: string
Expand Down
134 changes: 130 additions & 4 deletions config/v1/types_cluster_version.go
View file
Original file line number Diff line number Diff line change
Expand Up @@ -123,14 +123,26 @@ type ClusterVersionStatus struct {
// +optional
Conditions []ClusterOperatorStatusCondition `json:"conditions,omitempty"`

// availableUpdates contains the list of updates that are appropriate
// for this cluster. This list may be empty if no updates are recommended,
// if the update service is unavailable, or if an invalid channel has
// been specified.
// availableUpdates contains updates recommended for this
// cluster. Updates which appear in conditionalUpdates but not in
// availableUpdates may expose this cluster to known issues. This list
// may be empty if no updates are recommended, if the update service
// is unavailable, or if an invalid channel has been specified.
// +nullable
// +kubebuilder:validation:Required
// +required
AvailableUpdates []Release `json:"availableUpdates"`

// conditionalUpdates contains the list of updates that may be
// recommended for this cluster if it meets specific required
// conditions. Consumers interested in the set of updates that are
// actually recommended for this cluster should use
// availableUpdates. This list may be empty if no updates are
// recommended, if the update service is unavailable, or if an empty
// or invalid channel has been specified.
// +listType=atomic
// +optional
ConditionalUpdates []ConditionalUpdate `json:"conditionalUpdates,omitempty"`
}

// UpdateState is a constant representing whether an update was successfully
Expand Down Expand Up @@ -190,6 +202,14 @@ type UpdateHistory struct {
// +kubebuilder:validation:Required
// +required
Verified bool `json:"verified"`

// acceptedRisks records risks which were accepted to initiate the update.
// For example, it may menition an Upgradeable=False or missing signature
// that was overriden via desiredUpdate.force, or an update that was
// initiated despite not being in the availableUpdates set of recommended
// update targets.
// +optional
AcceptedRisks string `json:"acceptedRisks,omitempty"`
}

// ClusterID is string RFC4122 uuid.
Expand Down Expand Up @@ -290,6 +310,112 @@ type Release struct {
// availableUpdates field is accurate and recent.
const RetrievedUpdates ClusterStatusConditionType = "RetrievedUpdates"

// ConditionalUpdate represents an update which is recommended to some
// clusters on the version the current cluster is reconciling, but which
// may not be recommended for the current cluster.
type ConditionalUpdate struct {
// release is the target of the update.
// +kubebuilder:validation:Required
// +required
Release Release `json:"release"`

// risks represents the range of issues associated with
// updating to the target release. The cluster-version
// operator will evaluate all entries, and only recommend the
// update if there is at least one entry and all entries
// recommend the update.
// +kubebuilder:validation:Required
// +kubebuilder:validation:MinItems=1
// +patchMergeKey=name
// +patchStrategy=merge
// +listType=map
// +listMapKey=name
// +required
Risks []ConditionalUpdateRisk `json:"risks"`

// conditions represents the observations of the conditional update's
// current status. Known types are:
// * Evaluating, for whether the cluster-version operator will attempt to evaluate any risks[].matchingRules.
// * Recommended, for whether the update is recommended for the current cluster.
// +patchMergeKey=type
// +patchStrategy=merge
// +listType=map
// +listMapKey=type
Conditions []metav1.Condition `json:"conditions,omitempty" patchStrategy:"merge" patchMergeKey:"type" protobuf:"bytes,1,rep,name=conditions"`
}

// ConditionalUpdateRisk represents a reason and cluster-state
// for not recommending a conditional update.
// +k8s:deepcopy-gen=true
type ConditionalUpdateRisk struct {
// url contains information about this risk.
// +kubebuilder:validation:Required
// +kubebuilder:validation:Format=uri
// +kubebuilder:validation:MinLength=1
// +required
URL string `json:"url"`

// name is the CamelCase reason for not recommending a
// conditional update, in the event that matchingRules match the
// cluster state.
// +kubebuilder:validation:Required
// +kubebuilder:validation:MinLength=1
// +required
Name string `json:"name"`

// message provides additional information about the risk of
// updating, in the event that matchingRules match the cluster
// state. This is only to be consumed by humans. It may
// contain Line Feed characters (U+000A), which should be
// rendered as new lines.
// +kubebuilder:validation:Required
// +kubebuilder:validation:MinLength=1
// +required
Message string `json:"message"`

// matchingRules is a slice of conditions for deciding which
// clusters match the risk and which do not. The slice is
// ordered by decreasing precedence. The cluster-version
// operator will walk the slice in order, and stop after the
// first it can successfully evaluate. If no condition can be
// successfully evaluated, the update will not be recommended.
// +kubebuilder:validation:Required
// +kubebuilder:validation:MinItems=1
// +listType=atomic
// +required
MatchingRules []ClusterCondition `json:"matchingRules"`
}

// ClusterCondition is a union of typed cluster conditions. The 'type'
// property determines which of the type-specific properties are relevant.
// When evaluated on a cluster, the condition may match, not match, or
// fail to evaluate.
// +k8s:deepcopy-gen=true
type ClusterCondition struct {
// type represents the cluster-condition type. This defines
// the members and semantics of any additional properties.
// +kubebuilder:validation:Required
// +kubebuilder:validation:Enum={"Always","PromQL"}
// +required
Type string `json:"type"`

// promQL represents a cluster condition based on PromQL.
// +optional
PromQL *PromQLClusterCondition `json:"promql,omitempty"`
}

// PromQLClusterCondition represents a cluster condition based on PromQL.
type PromQLClusterCondition struct {
// PromQL is a PromQL query classifying clusters. This query
// query should return a 1 in the match case and a 0 in the
// does-not-match case case. Queries which return no time
// series, or which return values besides 0 or 1, are
// evaluation failures.
// +kubebuilder:validation:Required
// +required
PromQL string `json:"promql"`
}

// ClusterVersionList is a list of ClusterVersion resources.
//
// Compatibility level 1: Stable within a major release for a minimum of 12 months or 3 minor releases (whichever is longer).
Expand Down
Loading

0 comments on commit 03d33da

Please sign in to comment.