Prevent logs -f from repeating

The retryWatcher in client go seems to be broken and results
in the logs being restarted from the same spot each time despite
the version info being set.

Using a typical watch seems to fix the issue.

Signed-off-by: John Schnake <jschnake@vmware.com>

pkg/client/logs.go

@@ -29,7 +29,6 @@ import (
 	"k8s.io/apimachinery/pkg/watch"
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/tools/cache"
-	watchtool "k8s.io/client-go/tools/watch"
 )
 
 const (
@@ -188,17 +187,9 @@ func watchPodsToStreamLogs(client kubernetes.Interface, cfg *LogConfig, podCh ch
 		},
 	}
 
-	// You must get an initial resource version for the watcher; the retry watcher can't simply
-	// start at "now". It will err if given "", or "0" and API instructs users to not assume
-	// numerical or sequential access.
-	initVersionObj, err := client.CoreV1().Pods(cfg.Namespace).List(context.TODO(), listOptions)
+	watcher, err := lw.Watch(listOptions)
 	if err != nil {
-		return errors.Wrap(err, "failed to obtain initial resource version for retry watcher")
-	}
-
-	watcher, err := watchtool.NewRetryWatcher(initVersionObj.GetResourceVersion(), lw)
-	if err != nil {
-		return errors.Wrap(err, "failed to create retry watcher")
+		return errors.Wrap(err, "failed to start watching pod logs")
 	}
 	ch := watcher.ResultChan()