kubeedge cloud leader election

k8s.io/client-go/tools/leaderelection

- NewLeaderElector(leaderElectionConfig): leaderElector- LeaderElector.Run(ctx)- acquire- renew- tryAcquireOrRenew

TryToPatchPodReadinessGate

- Check environment \"CLOUDCORE_POD_NAME\

config.LeaderElection.LeaderElect

- Create recorder- New kube client- Create namespace if needed- Create leaderelection configuration- Set leaderelection callback

leaderelection

NewLeaderReadyzAdaptorRun

k8s.io/client-go/tools/record.broadcaster

- NewRecorder

k8s.io/client-go/tools/leaderelection/resourcelock

LeaderElectionRecord {  HolderIdentity        LeaseDurationSeconds  AcquireTime           RenewTime             LeaderTransitions    }

Run

- Try to patch PodReadinessGate if program runs in pod(status=false)- build leader election configuration- New leader electer- Set ready leader elector- leader elector run

NewCloudCoreCommand

registerModules

makeLeaderElectionConfig

- leaderelection:  LeaderElect: true  LeaseDuration: 15s  RenewDeadline: 10s  ResourceLock: endpointsleases  ResourceName: cloudcorelease  ResourceNamespace: kubeedge  RetryPeriod: 2s - Create resource lock- Create leaderelection configuration

httpserver

- DefaultCloudCoreReadyCheckURL = \"/readyz\"set the route \"GET\" handler as electorChecker

ReadyzAdaptor

timeoutLeaderElectorpointerLock

- Check: Is leader- SetLeaderElection- NewLeaderReadyzAdaptor