/
osdupgrader.go
171 lines (147 loc) · 6.95 KB
/
osdupgrader.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
package upgraders
import (
"context"
"time"
"github.com/go-logr/logr"
corev1 "k8s.io/api/core/v1"
"sigs.k8s.io/controller-runtime/pkg/client"
upgradev1alpha1 "github.com/openshift/managed-upgrade-operator/api/v1alpha1"
ac "github.com/openshift/managed-upgrade-operator/pkg/availabilitychecks"
cv "github.com/openshift/managed-upgrade-operator/pkg/clusterversion"
"github.com/openshift/managed-upgrade-operator/pkg/configmanager"
"github.com/openshift/managed-upgrade-operator/pkg/drain"
"github.com/openshift/managed-upgrade-operator/pkg/eventmanager"
"github.com/openshift/managed-upgrade-operator/pkg/machinery"
"github.com/openshift/managed-upgrade-operator/pkg/maintenance"
"github.com/openshift/managed-upgrade-operator/pkg/metrics"
"github.com/openshift/managed-upgrade-operator/pkg/notifier"
"github.com/openshift/managed-upgrade-operator/pkg/scaler"
"github.com/openshift/managed-upgrade-operator/pkg/upgradesteps"
)
// osdUpgrader is a cluster upgrader suitable for OpenShift Dedicated clusters.
// It inherits from the base clusterUpgrader.
type osdUpgrader struct {
*clusterUpgrader
}
// NewOSDUpgrader creates a new instance of an osdUpgrader
func NewOSDUpgrader(c client.Client, cfm configmanager.ConfigManager, mc metrics.Metrics, notifier eventmanager.EventManager) (*osdUpgrader, error) {
cfg := &upgraderConfig{}
err := cfm.Into(cfg)
if err != nil {
return nil, err
}
m, err := maintenance.NewBuilder().NewClient(c)
if err != nil {
return nil, err
}
acs, err := ac.GetAvailabilityCheckers(&cfg.ExtDependencyAvailabilityCheck)
if err != nil {
return nil, err
}
ou := osdUpgrader{
clusterUpgrader: &clusterUpgrader{
client: c,
metrics: mc,
cvClient: cv.NewCVClient(c),
notifier: notifier,
config: cfg,
scaler: scaler.NewScaler(),
drainstrategyBuilder: drain.NewBuilder(),
maintenance: m,
machinery: machinery.NewMachinery(),
availabilityCheckers: acs,
},
}
steps := []upgradesteps.UpgradeStep{
upgradesteps.Action(string(upgradev1alpha1.SendStartedNotification), ou.SendStartedNotification),
upgradesteps.Action(string(upgradev1alpha1.SendStartedNotification), ou.UpgradeDelayedCheck),
upgradesteps.Action(string(upgradev1alpha1.IsClusterUpgradable), ou.IsUpgradeable),
upgradesteps.Action(string(upgradev1alpha1.UpgradePreHealthCheck), ou.PreUpgradeHealthCheck),
upgradesteps.Action(string(upgradev1alpha1.ExtDepAvailabilityCheck), ou.ExternalDependencyAvailabilityCheck),
upgradesteps.Action(string(upgradev1alpha1.UpgradeScaleUpExtraNodes), ou.EnsureExtraUpgradeWorkers),
upgradesteps.Action(string(upgradev1alpha1.ControlPlaneMaintWindow), ou.CreateControlPlaneMaintWindow),
upgradesteps.Action(string(upgradev1alpha1.CommenceUpgrade), ou.CommenceUpgrade),
upgradesteps.Action(string(upgradev1alpha1.ControlPlaneUpgraded), ou.ControlPlaneUpgraded),
upgradesteps.Action(string(upgradev1alpha1.RemoveControlPlaneMaintWindow), ou.RemoveControlPlaneMaintWindow),
upgradesteps.Action(string(upgradev1alpha1.WorkersMaintWindow), ou.CreateWorkerMaintWindow),
upgradesteps.Action(string(upgradev1alpha1.AllWorkerNodesUpgraded), ou.AllWorkersUpgraded),
upgradesteps.Action(string(upgradev1alpha1.RemoveExtraScaledNodes), ou.RemoveExtraScaledNodes),
upgradesteps.Action(string(upgradev1alpha1.RemoveMaintWindow), ou.RemoveMaintWindow),
upgradesteps.Action(string(upgradev1alpha1.PostClusterHealthCheck), ou.PostUpgradeHealthCheck),
upgradesteps.Action(string(upgradev1alpha1.PostUpgradeProcedures), ou.PostUpgradeProcedures),
upgradesteps.Action(string(upgradev1alpha1.SendCompletedNotification), ou.SendCompletedNotification),
}
ou.steps = steps
return &ou, nil
}
// UpgradeCluster performs the upgrade of the cluster and returns an indication of the
// last-executed upgrade phase and any error associated with the phase execution.
//
// The UpgradeCluster enforces OSD policy around expiring upgrades if they do not commence
// within a given time period.
func (u *osdUpgrader) UpgradeCluster(ctx context.Context, upgradeConfig *upgradev1alpha1.UpgradeConfig, logger logr.Logger) (upgradev1alpha1.UpgradePhase, error) {
u.upgradeConfig = upgradeConfig
// OSD upgrader enforces a 'failure' policy if the upgrade does not commence within a time period
if cancelUpgrade, _ := shouldFailUpgrade(u.cvClient, u.config, u.upgradeConfig); cancelUpgrade {
return performUpgradeFailure(u.client, u.metrics, u.scaler, u.notifier, u.upgradeConfig, logger)
}
return u.runSteps(ctx, logger, u.steps)
}
// shouldFailUpgrade checks if the cluster has reached a condition during upgrade
// where it should be treated as failed.
// If the cluster should fail its upgrade a condition of 'true' is returned.
// Any error encountered in making this decision is returned.
func shouldFailUpgrade(cvClient cv.ClusterVersion, cfg *upgraderConfig, upgradeConfig *upgradev1alpha1.UpgradeConfig) (bool, error) {
commenced, err := cvClient.HasUpgradeCommenced(upgradeConfig)
if err != nil {
return false, err
}
// If the upgrade has commenced, there's no going back
if commenced {
return false, nil
}
// Get the managed upgrade start time from upgrade config history
h := upgradeConfig.Status.History.GetHistory(upgradeConfig.Spec.Desired.Version)
if h == nil {
return false, nil
}
startTime := h.StartTime.Time
upgradeWindowDuration := cfg.UpgradeWindow.GetUpgradeWindowTimeOutDuration()
if !startTime.IsZero() && upgradeWindowDuration > 0 && time.Now().After(startTime.Add(upgradeWindowDuration)) {
return true, nil
}
return false, nil
}
// performUpgradeFailure carries out routines related to moving to an upgrade-failed state
func performUpgradeFailure(c client.Client, metricsClient metrics.Metrics, s scaler.Scaler, nc eventmanager.EventManager, upgradeConfig *upgradev1alpha1.UpgradeConfig, logger logr.Logger) (upgradev1alpha1.UpgradePhase, error) {
// Set up return condition
h := upgradeConfig.Status.History.GetHistory(upgradeConfig.Spec.Desired.Version)
condition := &upgradev1alpha1.UpgradeCondition{
Type: "FailedUpgrade",
Status: corev1.ConditionFalse,
Reason: "Upgrade failed",
Message: "FailedUpgrade notification sent",
}
// TearDown the extra machineset
_, err := s.EnsureScaleDownNodes(c, nil, logger)
if err != nil {
logger.Error(err, "Failed to scale down the temporary upgrade machine when upgrade failed")
h.Conditions.SetCondition(*condition)
return h.Phase, nil
}
// Notify of failure
err = nc.Notify(notifier.MuoStateFailed)
if err != nil {
logger.Error(err, "Failed to notify of upgrade failure")
h.Conditions.SetCondition(*condition)
return h.Phase, nil
}
// flag window breached metric
metricsClient.UpdateMetricUpgradeWindowBreached(upgradeConfig.Name)
// cancel previously triggered metrics
metricsClient.ResetFailureMetrics()
// Update condition state to successful
condition.Status = corev1.ConditionTrue
h.Conditions.SetCondition(*condition)
return upgradev1alpha1.UpgradePhaseFailed, nil
}