Automatic merge from submit-queue
PLEG: reinspect pods that failed prior inspections
Fix the following sequence of events:
1. relist call 1 successfully inspects a pod (just has infra container)
1. relist call 2 gets an error inspecting the same pod (has infra container and a transient
container that failed to create) and doesn't update the old/new pod records
1. relist calls 3+ don't inspect the pod any more (just has infra container so it doesn't look like
anything changed)
This change adds a new list that keeps track of pods that failed inspection and retries them the
next time relist is called. Without this change, a pod in this state would never be inspected again,
its entry in the status cache would never be updated, and the pod worker would never call syncPod
again because the most recent entry in the status cache has an error associated with it. Without
this change, pods in this state would be stuck Terminating forever, unless the user issued a
deletion with a grace period value of 0.
Fixes#24819
cc @kubernetes/rh-cluster-infra @kubernetes/sig-node
Automatic merge from submit-queue
Make ThirdPartyResource a root scoped object
ThirdPartyResource (the registration of a third party type) belongs at the cluster scope. It results in resource handlers installed in every namespace, and the same name in two namespaces collides (namespace is ignored when determining group/kind).
ThirdPartyResourceData (an actual instance of that type) is still namespace-scoped.
This PR moves ThirdPartyResource to be a root scope object. Someone previously using ThirdPartyResource definitions in alpha should be able to move them from namespace to root scope like this:
setup (run on 1.2):
```
kubectl create ns ns1
echo '{"kind":"ThirdPartyResource","apiVersion":"extensions/v1beta1","metadata":{"name":"foo.example.com"},"versions":[{"name":"v8"}]}' | kubectl create -f - --namespace=ns1
echo '{"kind":"Foo","apiVersion":"example.com/v8","metadata":{"name":"MyFoo"},"testkey":"testvalue"}' | kubectl create -f - --namespace=ns1
```
export:
```
kubectl get thirdpartyresource --all-namespaces -o yaml > tprs.yaml
```
remove namespaced kind registrations (this shouldn't remove the data of that type, which is another possible issue):
```
kubectl delete -f tprs.yaml
```
... upgrade ...
re-register the custom types at the root scope:
```
kubectl create -f tprs.yaml
```
Additionally, pre-1.3 clients that expect to read/write ThirdPartyResource at a namespace scope will not be compatible with 1.3+ servers, and 1.3+ clients that expect to read/write ThirdPartyResource at a root scope will not be compatible with pre-1.3 servers.
Automatic merge from submit-queue
Define interfaces for kubelet pod admission and eviction
There is too much code and logic in `kubelet.go` that makes it hard to test functions in discrete pieces.
I propose an interface that an internal module can implement that will let it make an admission decision for a pod. If folks are ok with the pattern, I want to move the a) predicate checking, b) out of disk, c) eviction preventing best-effort pods being admitted into their own dedicated handlers that would be easier for us to mock test. We can then just write tests to ensure that the `Kubelet` calls a call-out, and we can write easier unit tests to ensure that dedicated handlers do the right thing.
The second interface I propose was a `PodEvictor` that is invoked in the main kubelet sync loop to know if pods should be pro-actively evicted from the machine. The current active deadline check should move into a simple evictor implementation, and I want to plug the out of resource killer code path as an implementation of the same interface.
@vishh @timothysc - if you guys can ack on this, I will add some unit testing to ensure we do the call-outs.
/cc @kubernetes/sig-node @kubernetes/rh-cluster-infra
Automatic merge from submit-queue
Avoid allocations and a reflect.Call in conversion
reflect.Call is fairly expensive, performing 8 allocations and having to
set up a call stack. Using a fairly straightforward to generate switch
statement, we can bypass that early in conversion (as long as the
function takes responsibility for invocation). We may also be able to
avoid an allocation for the conversion scope, but not positive yet.
```
benchmark old ns/op new ns/op delta
BenchmarkPodConversion-8 14713 12173 -17.26%
benchmark old allocs new allocs delta
BenchmarkPodConversion-8 80 72 -10.00%
benchmark old bytes new bytes delta
BenchmarkPodConversion-8 9133 8712 -4.61%
```
@wojtek-t related to #20309
Automatic merge from submit-queue
kubenet: fix up CNI bridge TX queue length if needed
CNI's bridge plugin mis-handles the TxQLen when creating the bridge,
leading to a zero-length TX queue. This doesn't typically cause
problems (since virtual interfaces don't have hard queue limits)
but when adding traffic shaping, some qdiscs pull their packet
limits from the TX queue length, leading to a packet limit of 0
in some cases. Until we can depend on a new enough version of
CNI, fix up the TX queue length internally.
Closes: https://github.com/kubernetes/kubernetes/issues/25092
Automatic merge from submit-queue
Remove nodeName from predicate signature.
With this approach, I'm getting the initial throughput (in empty cluster) in 1000-node cluster of ~95pods/s.
Which is ~30% improvement.
@kubernetes/sig-scalability
Automatic merge from submit-queue
Kubelet: Cleanup with new engine api
Finish step 2 of #23563
This PR:
1) Cleanup go-dockerclient reference in the code.
2) Bump up the engine-api version.
3) Cleanup the code with new engine-api.
Fixes#24076.
Fixes#23809.
/cc @yujuhong
Automatic merge from submit-queue
Delete pod with uid as precondition.
Addressed https://github.com/kubernetes/kubernetes/issues/25169#issuecomment-217033202.
Fix#25169Fix#24937
This PR change status manager to delete pods with uid as a precondition, so that kubelet won't delete pods with different uid but the same name and namespace accidentally.
/cc @yujuhong
Automatic merge from submit-queue
start etcd compactor in background
ref: #22448
What's in this PR?
- StartCompactor starts a compactor in the background in order to compact keys older than fixed time. We need to compact keys because we can't let on disk data grow forever. We save the most recent 10 minutes data. It should be enough for slow watchers and to tolerate burst. We might keep a longer history (12h) in the future once storage API can take advantage of multi-version key.
- Have only one compaction job for each cluster. Use endpoints from user input to differentiate clusters.
Automatic merge from submit-queue
API changes for Cascading deletion
This PR includes the necessary API changes to implement cascading deletion with finalizers as proposed is in #23656. Comments are welcome.
@lavalamp @derekwaynecarr @bgrant0607 @rata @hongchaodeng
Automatic merge from submit-queue
Allow etcd to store protobuf objects
Split storage serialization from client negotiation, and allow API server to take flag controlling serialization.
TODO:
* [x] API server still doesn't start - range allocation object doesn't seem to round trip correctly to etcd
* [ ] Verify that third party resources are ignoring protobuf (add a test)
* [ ] Add integration tests that verify storage is correctly protobuf
* [ ] Add a global default for which storage format to prefer?
