shim: retrieve oom events after starting sandbox

After starting the sandbox the shim will start polling for oom events from the agent. Propagates oom events through to containerd/cri. fixes kata-containers#2150 Signed-off-by: Alex Price <[email protected]>
awprice · Apr 28, 2020 · db28dcf · db28dcf
1 parent 86686b5
commit db28dcf
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 0 deletions.
diff --git a/containerd-shim-v2/start.go b/containerd-shim-v2/start.go
@@ -36,6 +36,9 @@ func startContainer(ctx context.Context, s *service, c *container) error {
 			return err
 		}
 		go watchSandbox(s)
+
+		// Start watching for oom events
+		go watchOOMEvents(s)
 	} else {
 		_, err := s.sandbox.StartContainer(c.id)
 		if err != nil {

diff --git a/containerd-shim-v2/wait.go b/containerd-shim-v2/wait.go
@@ -9,6 +9,7 @@ import (
 	"path"
 	"time"
 
+	"github.com/containerd/containerd/api/events"
 	"github.com/containerd/containerd/api/types/task"
 	"github.com/containerd/containerd/mount"
 	"github.com/sirupsen/logrus"
@@ -126,3 +127,26 @@ func watchSandbox(s *service) {
 	// Existing container/exec will be cleaned up by its waiters.
 	// No need to send async events here.
 }
+
+func watchOOMEvents(s *service) {
+	if s.sandbox == nil {
+		return
+	}
+
+	for {
+		select {
+		case <-s.ctx.Done():
+			return
+		default:
+			containerID, err := s.sandbox.GetOOMEvent()
+			if err != nil {
+				logrus.WithError(err).Warn("failed to get oom event from sandbox")
+				continue
+			}
+
+			s.send(&events.TaskOOM{
+				ContainerID: containerID,
+			})
+		}
+	}
+}