[Maintenance] Coordinator Health issue upgrade procedure (#1853)

ajanikow · web-flow · commit 8cd2cebb7c66 · 2025-03-25T16:27:19.000+01:00
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -13,6 +13,7 @@
 - (Feature) Emit Default values in Inventory
 - (Maintenance) Bump github.com/containerd/containerd to v1.7.27
 - (Feature) Rotation Order
+- (Maintenance) Coordinator Health issue upgrade procedure
 
 ## [1.2.46](https://github.com/arangodb/kube-arangodb/tree/1.2.46) (2025-02-24)
 - (Bugfix) Clean Phase change properly during upgrade
diff --git a/docs/integration-sidecar.md b/docs/integration-sidecar.md
@@ -1,9 +1,8 @@
 ---
 layout: page
-has_children: true
 title: Integration Sidecars
 parent: ArangoDBPlatform
-has_toc: false
+nav_order: 1
 ---
 
 # Integration 
diff --git a/docs/integration/authentication.v1.md b/docs/integration/authentication.v1.md
@@ -1,7 +1,7 @@
 ---
 layout: page
-title: Authentication V1
-parent: Integration Sidecars
+title: Integration Sidecar Authentication V1
+parent: ArangoDBPlatform
 ---
 
 # Authentication V1
diff --git a/docs/integration/authorization.v0.md b/docs/integration/authorization.v0.md
@@ -1,7 +1,7 @@
 ---
 layout: page
-title: Authorization V0
-parent: Integration Sidecars
+title: Integration Sidecar Authorization V0
+parent: ArangoDBPlatform
 ---
 
 # Authorization V0
diff --git a/docs/integration/scheduler.v1.md b/docs/integration/scheduler.v1.md
@@ -1,7 +1,7 @@
 ---
 layout: page
-title: Scheduler V1
-parent: Integration Sidecars
+title: Integration Sidecar Scheduler V1
+parent: ArangoDBPlatform
 ---
 
 # Scheduler V1
diff --git a/docs/integration/scheduler.v2.md b/docs/integration/scheduler.v2.md
@@ -1,7 +1,7 @@
 ---
 layout: page
-title: Scheduler V2
-parent: Integration Sidecars
+title: Integration Sidecar Scheduler V2
+parent: ArangoDBPlatform
 ---
 
 # Scheduler V2
diff --git a/docs/integration/shutdown.v1.md b/docs/integration/shutdown.v1.md
@@ -1,7 +1,7 @@
 ---
 layout: page
-title: Shutdown V1
-parent: Integration Sidecars
+title: Integration Sidecar Shutdown V1
+parent: ArangoDBPlatform
 ---
 
 # Shutdown V1
diff --git a/docs/integration/storage.v2.md b/docs/integration/storage.v2.md
@@ -1,7 +1,7 @@
 ---
 layout: page
-title: Storage V2
-parent: Integration Sidecars
+title: Integration Sidecar Storage V2
+parent: ArangoDBPlatform
 ---
 
 # Storage V2
diff --git a/docs/upgrade/health-endpoint-issues.md b/docs/upgrade/health-endpoint-issues.md
@@ -0,0 +1,14 @@
+---
+layout: page
+parent: Upgrading ArangoDB version
+title: Coordinator Health Endpoint Issue
+---
+
+# Coordinator Health Endpoint Issue
+
+Affected Versions:
+- < 3.12.4
+
+# Changes
+
+During the upgrade Operator will change default [Upgrade Order](../api/ArangoDeployment.V1.md#specupgradeorder) from `standard` to `coordinatorFirst` in order to update coordinators first.
diff --git a/pkg/deployment/reconcile/plan_builder_rotate_upgrade.go b/pkg/deployment/reconcile/plan_builder_rotate_upgrade.go
@@ -238,7 +238,8 @@ func (r *Reconciler) createUpgradePlanInternalCondition(apiObject k8sutil.APIObj
 }
 
 func (r *Reconciler) createUpgradePlanInternal(apiObject k8sutil.APIObject, spec api.DeploymentSpec, status api.DeploymentStatus, context PlanBuilderContext, decision updateUpgradeDecisionMap, agencyCache state.State) (api.Plan, bool) {
-	upgradeOrder := getUpgradeOrder(spec).Groups()
+	from, to := decision.GetFromToVersion()
+	upgradeOrder := getUpgradeOrder(spec, from, to).Groups()
 
 	for _, group := range upgradeOrder {
 		for _, m := range status.Members.AsListInGroup(group) {
@@ -610,7 +611,13 @@ func waitForMemberActions(group api.ServerGroup, member api.MemberStatus) api.Pl
 	}
 }
 
-func getUpgradeOrder(spec api.DeploymentSpec) api.DeploymentSpecOrder {
+func getUpgradeOrder(spec api.DeploymentSpec, from, to driver.Version) api.DeploymentSpecOrder {
+	if upgrade := spec.Upgrade; upgrade == nil || upgrade.Order == nil {
+		if to.CompareTo("3.12.4") >= 0 && from.CompareTo("3.12.4") < 0 && from.CompareTo("3.12.0") >= 0 && from != "" && to != "" {
+			return api.DeploymentSpecOrderCoordinatorFirst
+		}
+	}
+
 	return util.BoolSwitch(features.UpgradeAlternativeOrder().Enabled(), api.DeploymentSpecOrderCoordinatorFirst, spec.Upgrade.GetOrder(util.NewType(api.DeploymentSpecOrderStandard)))
 }
 
diff --git a/pkg/deployment/reconcile/plan_builder_rotate_upgrade_decision.go b/pkg/deployment/reconcile/plan_builder_rotate_upgrade_decision.go
@@ -21,6 +21,8 @@
 package reconcile
 
 import (
+	"github.com/arangodb/go-driver"
+
 	api "github.com/arangodb/kube-arangodb/pkg/apis/deployment/v1"
 	"github.com/arangodb/kube-arangodb/pkg/deployment/rotation"
 	"github.com/arangodb/kube-arangodb/pkg/util"
@@ -44,6 +46,24 @@ type updateUpgradeDecisionItem struct {
 
 type updateUpgradeDecisionMap map[updateUpgradeDecisionItem]updateUpgradeDecision
 
+func (u updateUpgradeDecisionMap) GetFromToVersion() (from driver.Version, to driver.Version) {
+	for _, el := range u {
+		if !el.upgradeDecision.UpgradeNeeded {
+			continue
+		}
+
+		if from == "" || from.CompareTo(el.upgradeDecision.FromVersion) > 0 {
+			from = el.upgradeDecision.FromVersion
+		}
+
+		if to == "" || to.CompareTo(el.upgradeDecision.ToVersion) < 0 {
+			to = el.upgradeDecision.ToVersion
+		}
+	}
+
+	return
+}
+
 func (u updateUpgradeDecisionMap) AreGroupsPendingUpgrade(groups ...api.ServerGroup) bool {
 	for _, group := range groups {
 		if u.IsGroupPendingUpgrade(group) {
diff --git a/pkg/deployment/reconcile/plan_builder_rotate_upgrade_test.go b/pkg/deployment/reconcile/plan_builder_rotate_upgrade_test.go
@@ -23,6 +23,7 @@ package reconcile
 import (
 	"testing"
 
+	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 
 	"github.com/arangodb/go-driver"
@@ -109,3 +110,14 @@ func Test_RotateUpgrade_Condition(t *testing.T) {
 		})
 	}
 }
+
+func Test_getUpgradeOrder(t *testing.T) {
+	assert.EqualValues(t, api.DeploymentSpecOrderStandard, getUpgradeOrder(api.DeploymentSpec{}, "", ""))
+	assert.EqualValues(t, api.DeploymentSpecOrderStandard, getUpgradeOrder(api.DeploymentSpec{}, "3.11.0", "3.12.4"))
+	assert.EqualValues(t, api.DeploymentSpecOrderCoordinatorFirst, getUpgradeOrder(api.DeploymentSpec{}, "3.12.0", "3.12.4"))
+	assert.EqualValues(t, api.DeploymentSpecOrderCoordinatorFirst, getUpgradeOrder(api.DeploymentSpec{}, "3.12.1", "3.12.4"))
+	assert.EqualValues(t, api.DeploymentSpecOrderCoordinatorFirst, getUpgradeOrder(api.DeploymentSpec{}, "3.12.2", "3.12.4"))
+	assert.EqualValues(t, api.DeploymentSpecOrderCoordinatorFirst, getUpgradeOrder(api.DeploymentSpec{}, "3.12.3", "3.12.4"))
+	assert.EqualValues(t, api.DeploymentSpecOrderStandard, getUpgradeOrder(api.DeploymentSpec{}, "3.12.4", "3.12.4"))
+	assert.EqualValues(t, api.DeploymentSpecOrderStandard, getUpgradeOrder(api.DeploymentSpec{}, "3.12.4", "3.12.5"))
+}

Original file line number	Diff line number	Diff line change
`@@ -238,7 +238,8 @@ func (r *Reconciler) createUpgradePlanInternalCondition(apiObject k8sutil.APIObj`
`238`	`238`	`}`
`239`	`239`
`240`	`240`	`func (r *Reconciler) createUpgradePlanInternal(apiObject k8sutil.APIObject, spec api.DeploymentSpec, status api.DeploymentStatus, context PlanBuilderContext, decision updateUpgradeDecisionMap, agencyCache state.State) (api.Plan, bool) {`
`241`		`- upgradeOrder := getUpgradeOrder(spec).Groups()`
	`241`	`+ from, to := decision.GetFromToVersion()`
	`242`	`+ upgradeOrder := getUpgradeOrder(spec, from, to).Groups()`
`242`	`243`
`243`	`244`	`for _, group := range upgradeOrder {`
`244`	`245`	`for _, m := range status.Members.AsListInGroup(group) {`
`@@ -610,7 +611,13 @@ func waitForMemberActions(group api.ServerGroup, member api.MemberStatus) api.Pl`
`610`	`611`	`}`
`611`	`612`	`}`
`612`	`613`
`613`		`-func getUpgradeOrder(spec api.DeploymentSpec) api.DeploymentSpecOrder {`
	`614`	`+func getUpgradeOrder(spec api.DeploymentSpec, from, to driver.Version) api.DeploymentSpecOrder {`
	`615`	`+ if upgrade := spec.Upgrade; upgrade == nil \|\| upgrade.Order == nil {`
	`616`	`+ if to.CompareTo("3.12.4") >= 0 && from.CompareTo("3.12.4") < 0 && from.CompareTo("3.12.0") >= 0 && from != "" && to != "" {`
	`617`	`+ return api.DeploymentSpecOrderCoordinatorFirst`
	`618`	`+ }`
	`619`	`+ }`
	`620`	`+`
`614`	`621`	`return util.BoolSwitch(features.UpgradeAlternativeOrder().Enabled(), api.DeploymentSpecOrderCoordinatorFirst, spec.Upgrade.GetOrder(util.NewType(api.DeploymentSpecOrderStandard)))`
`615`	`622`	`}`
`616`	`623`