data mover ms node agent resume

Signed-off-by: Lyndon-Li <lyonghui@vmware.com>
This commit is contained in:
Lyndon-Li
2024-08-06 16:25:56 +08:00
parent d4e743b138
commit a523d10802
9 changed files with 806 additions and 509 deletions
+14 -29
View File
@@ -292,18 +292,28 @@ func (s *nodeAgentServer) run() {
if s.dataPathConfigs != nil && len(s.dataPathConfigs.LoadAffinity) > 0 {
loadAffinity = s.dataPathConfigs.LoadAffinity[0]
}
dataUploadReconciler := controller.NewDataUploadReconciler(s.mgr.GetClient(), s.kubeClient, s.csiSnapshotClient.SnapshotV1(), s.dataPathMgr, loadAffinity, repoEnsurer, clock.RealClock{}, credentialGetter, s.nodeName, s.fileSystem, s.config.dataMoverPrepareTimeout, s.logger, s.metrics)
s.attemptDataUploadResume(dataUploadReconciler)
dataUploadReconciler := controller.NewDataUploadReconciler(s.mgr.GetClient(), s.mgr, s.kubeClient, s.csiSnapshotClient.SnapshotV1(), s.dataPathMgr, loadAffinity, repoEnsurer, clock.RealClock{}, credentialGetter, s.nodeName, s.fileSystem, s.config.dataMoverPrepareTimeout, s.logger, s.metrics)
if err = dataUploadReconciler.SetupWithManager(s.mgr); err != nil {
s.logger.WithError(err).Fatal("Unable to create the data upload controller")
}
dataDownloadReconciler := controller.NewDataDownloadReconciler(s.mgr.GetClient(), s.kubeClient, s.dataPathMgr, repoEnsurer, credentialGetter, s.nodeName, s.config.dataMoverPrepareTimeout, s.logger, s.metrics)
s.attemptDataDownloadResume(dataDownloadReconciler)
dataDownloadReconciler := controller.NewDataDownloadReconciler(s.mgr.GetClient(), s.mgr, s.kubeClient, s.dataPathMgr, repoEnsurer, credentialGetter, s.nodeName, s.config.dataMoverPrepareTimeout, s.logger, s.metrics)
if err = dataDownloadReconciler.SetupWithManager(s.mgr); err != nil {
s.logger.WithError(err).Fatal("Unable to create the data download controller")
}
go func() {
s.mgr.GetCache().WaitForCacheSync(s.ctx)
if err := dataUploadReconciler.AttemptDataUploadResume(s.ctx, s.mgr.GetClient(), s.logger.WithField("node", s.nodeName), s.namespace); err != nil {
s.logger.WithError(errors.WithStack(err)).Error("failed to attempt data upload resume")
}
if err := dataDownloadReconciler.AttemptDataDownloadResume(s.ctx, s.mgr.GetClient(), s.logger.WithField("node", s.nodeName), s.namespace); err != nil {
s.logger.WithError(errors.WithStack(err)).Error("failed to attempt data download resume")
}
}()
s.logger.Info("Controllers starting...")
if err := s.mgr.Start(ctrl.SetupSignalHandler()); err != nil {
@@ -373,31 +383,6 @@ func (s *nodeAgentServer) markInProgressCRsFailed() {
s.markInProgressPVRsFailed(client)
}
func (s *nodeAgentServer) attemptDataUploadResume(r *controller.DataUploadReconciler) {
// the function is called before starting the controller manager, the embedded client isn't ready to use, so create a new one here
client, err := ctrlclient.New(s.mgr.GetConfig(), ctrlclient.Options{Scheme: s.mgr.GetScheme()})
if err != nil {
s.logger.WithError(errors.WithStack(err)).Error("failed to create client")
return
}
if err := r.AttemptDataUploadResume(s.ctx, client, s.logger.WithField("node", s.nodeName), s.namespace); err != nil {
s.logger.WithError(errors.WithStack(err)).Error("failed to attempt data upload resume")
}
}
func (s *nodeAgentServer) attemptDataDownloadResume(r *controller.DataDownloadReconciler) {
// the function is called before starting the controller manager, the embedded client isn't ready to use, so create a new one here
client, err := ctrlclient.New(s.mgr.GetConfig(), ctrlclient.Options{Scheme: s.mgr.GetScheme()})
if err != nil {
s.logger.WithError(errors.WithStack(err)).Error("failed to create client")
return
}
if err := r.AttemptDataDownloadResume(s.ctx, client, s.logger.WithField("node", s.nodeName), s.namespace); err != nil {
s.logger.WithError(errors.WithStack(err)).Error("failed to attempt data download resume")
}
}
func (s *nodeAgentServer) markInProgressPVBsFailed(client ctrlclient.Client) {
pvbs := &velerov1api.PodVolumeBackupList{}
if err := client.List(s.ctx, pvbs, &ctrlclient.ListOptions{Namespace: s.namespace}); err != nil {