github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2025-08-07 03:03:59 +00:00
Code Issues Projects Releases Wiki Activity

Fix issue in enabling evented pleg feature gate

Browse Source 
Fixes https://github.com/kubernetes/kubernetes/issues/120941
GetNewerThan() call isn't blocking until the pod status/cache is updated and returning the empty pod status.
Hence, whenever the `SyncLoop ADD/UPDATE/RECONCILE` functions are called multiple times in a very less time interval,
Kubelet calls multiple `CreateContainer` CRI api that results in the creation of duplicate containers within a given pod.
The initially created conainer keeps `Running` and the later container keeps `Exiting` and hence resulting the pod in `CrashLoopBackOff` state forever

Signed-off-by: Sai Ramesh Vanka <svanka@redhat.com>
This commit is contained in:
Sai Ramesh Vanka 2023-10-17 13:07:01 +05:30
parent afe4c041f3
commit 1715fc0ca0
1 changed files with 0 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
pkg/kubelet/container/cache.go
View File

@ -157,29 +157,6 @@ func (c *cache) get(id types.UID) *data {
// Otherwise, it returns nil. The caller should acquire the lock.
func (c *cache) getIfNewerThan(id types.UID, minTime time.Time) *data {
d, ok := c.pods[id]
if utilfeature.DefaultFeatureGate.Enabled(features.EventedPLEG) {
// Evented PLEG has CREATED, STARTED, STOPPED and DELETED events
// However if the container creation fails for some reason there is no
// CRI event received by the kubelet and that pod will get stuck a
// GetNewerThan call in the pod workers. This is reproducible with
// the node e2e test,
// https://github.com/kubernetes/kubernetes/blob/83415e5c9e6e59a3d60a148160490560af2178a1/test/e2e_node/pod_hostnamefqdn_test.go#L161
// which forces failure during pod creation. This issue also exists in
// Generic PLEG but since it updates global timestamp periodically
// the GetNewerThan call gets unstuck.
// During node e2e tests, it was observed this change does not have any
// adverse impact on the behaviour of the Generic PLEG as well.
switch {
case !ok:
return makeDefaultData(id)
case ok && (d.modified.After(minTime) || (c.timestamp != nil && c.timestamp.After(minTime))):
return d
default:
return nil
}
}
globalTimestampIsNewer := (c.timestamp != nil && c.timestamp.After(minTime))
if !ok && globalTimestampIsNewer {
// Status is not cached, but the global timestamp is newer than