/
upgrade.go
365 lines (332 loc) · 12 KB
/
upgrade.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
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
package reconcilers
import (
"context"
"errors"
"fmt"
"time"
"github.com/Masterminds/semver"
"github.com/banzaicloud/operator-tools/pkg/reconciler"
appsv1 "k8s.io/api/apps/v1"
corev1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/types"
"k8s.io/client-go/tools/record"
"k8s.io/client-go/util/retry"
opsterv1 "opensearch.opster.io/api/v1"
"opensearch.opster.io/opensearch-gateway/services"
"opensearch.opster.io/pkg/builders"
"opensearch.opster.io/pkg/helpers"
"opensearch.opster.io/pkg/reconcilers/util"
ctrl "sigs.k8s.io/controller-runtime"
"sigs.k8s.io/controller-runtime/pkg/client"
"sigs.k8s.io/controller-runtime/pkg/log"
)
var (
ErrVersionDowngrade = errors.New("version requested is downgrade")
ErrMajorVersionJump = errors.New("version request is more than 1 major version ahead")
ErrUnexpectedStatus = errors.New("unexpected upgrade status")
)
type UpgradeReconciler struct {
client.Client
reconciler.ResourceReconciler
ctx context.Context
osClient *services.OsClusterClient
recorder record.EventRecorder
reconcilerContext *ReconcilerContext
instance *opsterv1.OpenSearchCluster
}
func NewUpgradeReconciler(
client client.Client,
ctx context.Context,
recorder record.EventRecorder,
reconcilerContext *ReconcilerContext,
instance *opsterv1.OpenSearchCluster,
opts ...reconciler.ResourceReconcilerOption,
) *UpgradeReconciler {
return &UpgradeReconciler{
Client: client,
ResourceReconciler: reconciler.NewReconcilerWith(client,
append(opts, reconciler.WithLog(log.FromContext(ctx).WithValues("reconciler", "upgrade")))...),
ctx: ctx,
recorder: recorder,
reconcilerContext: reconcilerContext,
instance: instance,
}
}
func (r *UpgradeReconciler) Reconcile() (ctrl.Result, error) {
// If versions are in sync do nothing
if r.instance.Spec.General.Version == r.instance.Status.Version {
return ctrl.Result{}, nil
}
annotations := map[string]string{"cluster-name": r.instance.GetName()}
lg := log.FromContext(r.ctx)
// If version validation fails log a warning and do nothing
if err := r.validateUpgrade(); err != nil {
lg.V(1).Error(err, "version validation failed", "currentVersion", r.instance.Status.Version, "requestedVersion", r.instance.Spec.General.Version)
r.recorder.AnnotatedEventf(r.instance, annotations, "Normal", "Upgrade", "Failed to validation version, currentVersion: %s , requestedVersion: %s", r.instance.Status.Version, r.instance.Spec.General.Version)
return ctrl.Result{}, err
}
var err error
r.osClient, err = util.CreateClientForCluster(r.ctx, r.Client, r.instance, nil)
if err != nil {
return ctrl.Result{}, err
}
//Fetch the working nodepool
nodePool, currentStatus := r.findWorkingNodePool()
// Work on the current nodepool as appropriate
switch currentStatus.Status {
case "Pending":
// Set it to upgrading and requeue
err := retry.RetryOnConflict(retry.DefaultRetry, func() error {
if err := r.Get(r.ctx, client.ObjectKeyFromObject(r.instance), r.instance); err != nil {
return err
}
currentStatus.Status = "Upgrading"
r.instance.Status.ComponentsStatus = append(r.instance.Status.ComponentsStatus, currentStatus)
return r.Status().Update(r.ctx, r.instance)
})
r.recorder.AnnotatedEventf(r.instance, annotations, "Normal", "Upgrade", "Start to upgrade of data node pool %s", currentStatus.Component)
return ctrl.Result{
Requeue: true,
RequeueAfter: 15 * time.Second,
}, err
case "NonDataPending":
// Set it to upgrading and requeue
err := retry.RetryOnConflict(retry.DefaultRetry, func() error {
if err := r.Get(r.ctx, client.ObjectKeyFromObject(r.instance), r.instance); err != nil {
return err
}
currentStatus.Status = "UntrackedUpgrade"
r.instance.Status.ComponentsStatus = append(r.instance.Status.ComponentsStatus, currentStatus)
return r.Status().Update(r.ctx, r.instance)
})
r.recorder.AnnotatedEventf(r.instance, annotations, "Normal", "Upgrade", "Start to rollout of non data node pool %s", currentStatus.Component)
return ctrl.Result{
Requeue: true,
RequeueAfter: 15 * time.Second,
}, err
case "Upgrading":
err := r.doDataNodeUpgrade(nodePool)
return ctrl.Result{
Requeue: true,
RequeueAfter: 30 * time.Second,
}, err
case "Finished":
// Cleanup status after successful upgrade
err := retry.RetryOnConflict(retry.DefaultRetry, func() error {
if err := r.Get(r.ctx, client.ObjectKeyFromObject(r.instance), r.instance); err != nil {
return err
}
r.instance.Status.Version = r.instance.Spec.General.Version
for _, pool := range r.instance.Spec.NodePools {
componentStatus := opsterv1.ComponentStatus{
Component: "Upgrader",
Description: pool.Component,
}
currentStatus, found := helpers.FindFirstPartial(r.instance.Status.ComponentsStatus, componentStatus, helpers.GetByDescriptionAndGroup)
if found {
r.instance.Status.ComponentsStatus = helpers.RemoveIt(currentStatus, r.instance.Status.ComponentsStatus)
}
}
return r.Status().Update(r.ctx, r.instance)
})
r.recorder.AnnotatedEventf(r.instance, annotations, "Normal", "Upgrade", "Finished to upgrade - NewVersion: %s", r.instance.Status.Version)
return ctrl.Result{}, err
default:
// We should never get here so return an error
return ctrl.Result{}, ErrUnexpectedStatus
}
}
// Currently provides basic validation on versions.
// TODO Improve the validation (maybe allow patch version downgrades)
func (r *UpgradeReconciler) validateUpgrade() error {
// Parse versions
existing, err := semver.NewVersion(r.instance.Status.Version)
if err != nil {
return err
}
new, err := semver.NewVersion(r.instance.Spec.General.Version)
if err != nil {
return err
}
annotations := map[string]string{"cluster-name": r.instance.GetName()}
// Don't allow version downgrades as they might cause unexpected issues
if new.LessThan(existing) {
r.recorder.AnnotatedEventf(r.instance, annotations, "Warning", "Upgrade", " Notice - invalid version - specified version is more than 1 major version greater than existing")
return ErrVersionDowngrade
}
// Don't allow more than one major version upgrade
nextMajor := existing.IncMajor().IncMajor()
upgradeConstraint, err := semver.NewConstraint(fmt.Sprintf("< %s", nextMajor.String()))
if err != nil {
return err
}
if !upgradeConstraint.Check(new) {
r.recorder.AnnotatedEventf(r.instance, annotations, "Warning", "invalid version", "specified version is more than 1 major version greater than existing")
return ErrMajorVersionJump
}
return nil
}
// Find which nodepool to work on
func (r *UpgradeReconciler) findWorkingNodePool() (opsterv1.NodePool, opsterv1.ComponentStatus) {
// First sort node pools
var dataNodes, dataAndMasterNodes, otherNodes []opsterv1.NodePool
annotations := map[string]string{"cluster-name": r.instance.GetName()}
for _, nodePool := range r.instance.Spec.NodePools {
if helpers.ContainsString(nodePool.Roles, "data") {
if helpers.ContainsString(nodePool.Roles, "master") {
dataAndMasterNodes = append(dataAndMasterNodes, nodePool)
} else {
dataNodes = append(dataNodes, nodePool)
}
} else {
otherNodes = append(otherNodes, nodePool)
}
}
// First work on data only nodes
// Complete the in progress node first
pool, found := r.findInProgress(dataNodes)
if found {
return pool, opsterv1.ComponentStatus{
Component: "Upgrader",
Description: pool.Component,
Status: "Upgrading",
}
}
// Pick the first unworked on node next
pool, found = r.findNextPool(dataNodes)
if found {
return pool, opsterv1.ComponentStatus{
Component: "Upgrader",
Description: pool.Component,
Status: "Pending",
}
}
// Next do the same for any nodes that are data and master
pool, found = r.findInProgress(dataAndMasterNodes)
if found {
return pool, opsterv1.ComponentStatus{
Component: "Upgrader",
Description: pool.Component,
Status: "Upgrading",
}
}
pool, found = r.findNextPool(dataAndMasterNodes)
if found {
return pool, opsterv1.ComponentStatus{
Component: "Upgrader",
Description: pool.Component,
Status: "Pending",
}
}
// Finally do the non data nodes. We can let the kubernetes rollout logic do the work here
pool, found = r.findNextPool(otherNodes)
if found {
return pool, opsterv1.ComponentStatus{
Component: "Upgrader",
Description: pool.Component,
Status: "NonDataPending",
}
}
// If we get here all nodes should be upgraded
r.recorder.AnnotatedEventf(r.instance, annotations, "Normal", "Upgrade", "Finished to upgrade - NewVersion: %s", r.instance.Status.Version)
return opsterv1.NodePool{}, opsterv1.ComponentStatus{
Component: "Upgrade",
Status: "Finished",
}
}
func (r *UpgradeReconciler) findInProgress(pools []opsterv1.NodePool) (opsterv1.NodePool, bool) {
for _, nodePool := range pools {
componentStatus := opsterv1.ComponentStatus{
Component: "Upgrader",
Description: nodePool.Component,
}
currentStatus, found := helpers.FindFirstPartial(r.instance.Status.ComponentsStatus, componentStatus, helpers.GetByDescriptionAndGroup)
if found && currentStatus.Status == "Upgrading" {
return nodePool, true
}
}
return opsterv1.NodePool{}, false
}
func (r *UpgradeReconciler) findNextPool(pools []opsterv1.NodePool) (opsterv1.NodePool, bool) {
for _, nodePool := range pools {
componentStatus := opsterv1.ComponentStatus{
Component: "Upgrader",
Description: nodePool.Component,
}
_, found := helpers.FindFirstPartial(r.instance.Status.ComponentsStatus, componentStatus, helpers.GetByDescriptionAndGroup)
if !found {
return nodePool, true
}
}
return opsterv1.NodePool{}, false
}
func (r *UpgradeReconciler) doDataNodeUpgrade(pool opsterv1.NodePool) error {
// Fetch the STS
lg := log.FromContext(r.ctx).WithValues("reconciler", "upgrader")
stsName := builders.StsName(r.instance, &pool)
sts := &appsv1.StatefulSet{}
annotations := map[string]string{"cluster-name": r.instance.GetName()}
if err := r.Get(r.ctx, types.NamespacedName{
Name: stsName,
Namespace: r.instance.Namespace,
}, sts); err != nil {
return err
}
dataCount := builders.DataNodesCount(r.ctx, r.Client, r.instance)
if dataCount == 2 && r.instance.Spec.General.DrainDataNodes {
lg.Info("only 2 data nodes and drain is set, some shards may not drain")
}
ready, err := services.CheckClusterStatusForRestart(r.osClient, r.instance.Spec.General.DrainDataNodes)
if err != nil {
return err
}
if !ready {
return nil
}
// Work around for https://github.com/kubernetes/kubernetes/issues/73492
// If upgrade on this node pool is complete update status and return
if sts.Status.UpdatedReplicas == sts.Status.Replicas {
return retry.RetryOnConflict(retry.DefaultRetry, func() error {
if err := r.Get(r.ctx, client.ObjectKeyFromObject(r.instance), r.instance); err != nil {
return err
}
currentStatus := opsterv1.ComponentStatus{
Component: "Upgrader",
Status: "Upgrading",
Description: pool.Component,
}
componentStatus := opsterv1.ComponentStatus{
Component: "Upgrader",
Status: "Upgraded",
Description: pool.Component,
}
r.instance.Status.ComponentsStatus = helpers.Replace(currentStatus, componentStatus, r.instance.Status.ComponentsStatus)
r.recorder.AnnotatedEventf(r.instance, annotations, "Normal", "Upgrade", "Finished to upgrade data node pool %s", currentStatus.Component)
return r.Status().Update(r.ctx, r.instance)
})
}
workingPod := builders.WorkingPodForRollingRestart(sts)
ready, err = services.PreparePodForDelete(r.osClient, workingPod, r.instance.Spec.General.DrainDataNodes, dataCount)
if err != nil {
return err
}
if !ready {
return nil
}
err = r.Delete(r.ctx, &corev1.Pod{
ObjectMeta: metav1.ObjectMeta{
Name: workingPod,
Namespace: sts.Namespace,
},
})
if err != nil {
return err
}
// If we are draining nodes remove the exclusion after the pod is deleted
if r.instance.Spec.General.DrainDataNodes {
_, err = services.RemoveExcludeNodeHost(r.osClient, workingPod)
return err
}
return nil
}