Automatic merge from submit-queue (batch tested with PRs 38318, 38258)
kernel memcg notification enabled via experimental flag
Kubelet integrates with kernel memcg notification API if and only if enabled via experimental flag.
Automatic merge from submit-queue
add a configuration for kubelet to register as a node with taints
and deprecate --register-schedulable
ref #28687#29178
cc @dchen1107 @davidopp @roberthbailey
Automatic merge from submit-queue (batch tested with PRs 37208, 37446, 37420)
Kubelet log modification
Keep in line with the other error logs in the function.
After return, the caller records the error log.Delete redundant logs
Automatic merge from submit-queue
Clean up redundant tests in image_manager_test
There was a lot of overlap between parallel and serialized puller tests,
extracted most of these tests internals to separate functions.
Automatic merge from submit-queue
Function annotation modification
“return kl.pleg.Healthy()”，Based on the return function,"healty" to "healthy" better
Automatic merge from submit-queue
Keep host port socket open for kubenet
fixes#37087
**Release note**:
<!-- Steps to write your release note:
1. Use the release-note-* labels to set the release note state (if you have access)
2. Enter your extended release note in the below block; leaving it blank means using the PR title as the release note. If no release note is required, just write `NONE`.
-->
```release-note
NONE
```
When cni is set to kubenet, kubelet should hold the host port socket,
so that other application in this node could not listen/bind this port
any more. However, the sockets are closed accidentally, because
kubelet forget to reconcile the protocol format before comparing.
@kubernetes/sig-network
Automatic merge from submit-queue
RunnningContainerStatues spelling mistake
runtime.go：in the function GetRunningContainerStatuses， runnningContainerStatues spelling mistake, modified into runningContainerStatus
Automatic merge from submit-queue
kubelet: don't reject pods without adding them to the pod manager
kubelet relies on the pod manager as a cache of the pods in the apiserver (and
other sources) . The cache should be kept up-to-date even when rejecting pods.
Without this, kubelet may decide at any point to drop the status update
(request to the apiserver) for the rejected pod since it would think the pod no
longer exists in the apiserver.
This should fix#37658
Automatic merge from submit-queue
remove checking mount point in cleanupOrphanedPodDirs
To avoid nfs hung problem, remove the mountpoint checking code in
cleanupOrphanedPodDirs(). This removal should still be safe because it checks whether there are still directories under pod's volume and if so, do not delete the pod directory.
Note: After removing the mountpoint check code in cleanupOrphanedPodDirs(), the directories might not be cleaned up in such situation.
1. delete pod, kubelet reconciler tries to unmount the volume directory successfully
2. before reconciler tries to delete the volume directory, kubelet gets retarted
3. since under pod directory, there are still volume directors exist (but not mounted), cleanupOrphanedPodDIrs() will not clean them up.
Will work on a follow up PR to solve above issue.
kubelet relies on the pod manager as a cache of the pods in the apiserver (and
other sources) . The cache should be kept up-to-date even when rejecting pods.
Without this, kubelet may decide at any point to drop the status update
(request to the apiserver) for the rejected pod since it would think the pod no
longer exists in the apiserver.
Also check if the pod to-be-admitted has terminated or not. In the case where
it has terminated, skip the admission process completely.
Automatic merge from submit-queue
Fix package aliases to follow golang convention
Some package aliases are not not align with golang convention https://blog.golang.org/package-names. This PR fixes them. Also adds a verify script and presubmit checks.
Fixes#35070.
cc/ @timstclair @Random-Liu
When cni is set to kubenet, kubelet should hold the host port socket,
so that other application in this node could not listen/bind this port
any more. However, the sockets are closed accidentally, because
kubelet forget to reconcile the protocol format before comparing.
Automatic merge from submit-queue
kubelet: eviction: add memcg threshold notifier to improve eviction responsiveness
This PR adds the ability for the eviction code to get immediate notification from the kernel when the available memory in the root cgroup falls below a user defined threshold, controlled by setting the `memory.available` siginal with the `--eviction-hard` flag.
This PR by itself, doesn't change anything as the frequency at which new stats can be obtained is currently controlled by the cadvisor housekeeping interval. That being the case, the call to `synchronize()` by the notification loop will very likely get stale stats and not act any more quickly than it does now.
However, whenever cadvisor does get on-demand stat gathering ability, this will improve eviction responsiveness by getting async notification of the root cgroup memory state rather than relying on polling cadvisor.
@vishh @derekwaynecarr @kubernetes/rh-cluster-infra
Automatic merge from submit-queue
fix leaking memory backed volumes of terminated pods
Currently, we allow volumes to remain mounted on the node, even though the pod is terminated. This creates a vector for a malicious user to exhaust memory on the node by creating memory backed volumes containing large files.
This PR removes memory backed volumes (emptyDir w/ medium Memory, secrets, configmaps) of terminated pods from the node.
@saad-ali @derekwaynecarr
Automatic merge from submit-queue
Fix hostname truncate.
Fixes https://github.com/kubernetes/kubernetes/issues/36951.
This PR will keep truncating the hostname until the ending character is valid.
/cc @kubernetes/sig-node
Mark v1.5 because this is a bug fix.
/cc @saad-ali
We have observed that, after failing to create a container due to "device or
resource busy", docker may end up having inconsistent internal state. One
symptom is that docker will not report the existence of the "failed to create"
container, but if kubelet tries to create a new container with the same name,
docker will error out with a naming conflict message.
To work around this, this commit parses the creation error message and if there
is a naming conflict, it would attempt to remove the existing container.
Automatic merge from submit-queue
CRI: add docs for sysctls
#34830 adds `sysctls` features in CRI, it is based on sandbox annotations, this PR adds docs for it.
@yujuhong @timstclair @jonboulle
Automatic merge from submit-queue
CRI: Clarify User in CRI.
Addressed https://github.com/kubernetes/kubernetes/pull/36423#issuecomment-259343135.
This PR clarifies the user related fields in CRI.
One question is that:
What is the meaning of the `run_as_user` field in `LinuxSandboxSecurityContext`?
* **Is it user on the host?** Then it doesn't make sense, user shouldn't care about what users are on the host.
* **Is it user inside the infra container image?** This is how the field is currently used. However, Infra container is docker specific, I'm not sure whether we should expose this in CRI.
* **Is it the default user inside the pod?** It tells runtime that if there is a container (infra container, or some other helper containers like streaming container etc.), if their `user` is not specified, use the default "sandbox user". Then how can we guarantee that infra or helper container image have the `user`?
* **It doesn't make sense?** If we remove it, we are relying on the shim to set right user (maybe always root) for infra or helper containers (if there will be any in the future), I'm not sure whether this is what we expect.
@yujuhong @feiskyer @jonboulle @yifan-gu
/cc @kubernetes/sig-node
Automatic merge from submit-queue
[kubelet] rename --cgroups-per-qos to --experimental-cgroups-per-qos
This reflects the true nature of "cgroups per qos" feature.
```release-note
* Rename `--cgroups-per-qos` to `--experimental-cgroups-per-qos` in Kubelet
```
Automatic merge from submit-queue
fix issue in reconstruct volume data when kubelet restarts
During state reconstruction when kubelet restarts, outerVolueSpecName
cannot be recovered by scanning the disk directories. But this
information is used by volume manager to check whether pod's volume is
mounted or not. There are two possible cases:
1. pod is not deleted during kubelet restarts so that desired state
should have the information. reconciler.updateState() will use this
inforamtion to update.
2. pod is deleted during this period, reconciler has to use
InnerVolumeSpecName, but it should be ok since this information will not
be used for volume cleanup (umount)
Automatic merge from submit-queue
CRI: general grammar/spelling/consistency cleanup
No semantic changes, but a lot of shuffling of docstrings to make things
more consistent. In particular, standardise on the zeroth-article (i.e.
prefer `// Version` to `// The version`) and ending all docstrings with
periods.
(This knowingly conflicts with #36446 and intentionally omits changing the
Annotations field - I'll rebase this or that respectively as necessary.)
During state reconstruction when kubelet restarts, outerVolueSpecName
cannot be recovered by scanning the disk directories. But this
information is used by volume manager to check whether pod's volume is
mounted or not. There are two possible cases:
1. pod is not deleted during kubelet restarts so that desired state
should have the information. reconciler.updateState() will use this
inforamtion to update.
2. pod is deleted during this period, reconciler has to use
InnerVolumeSpecName, but it should be ok since this information will not
be used for volume cleanup (umount)
Automatic merge from submit-queue
Fix getting cgroup pids
Fixes https://github.com/kubernetes/kubernetes/issues/35214, https://github.com/kubernetes/kubernetes/issues/33232
Verified manually, but I didn't have time to run all the e2e's yet (will check it in the morning).
This should be cherry-picked into 1.4, and merged into 1.5 (/cc @saad-ali )
```release-note
Fix fetching pids running in a cgroup, which caused problems with OOM score adjustments & setting the /system cgroup ("misc" in the summary API).
```
/cc @kubernetes/sig-node
Automatic merge from submit-queue
Add e2e node test for log path
fixes#34661
A node e2e test to check if container logs files are properly created with right content.
Since the log files under `/var/log/containers` are actually symbolic of docker containers log files, we can not use a pod to mount them in and do check (symbolic doesn't supported by docker volume).
cc @Random-Liu
Automatic merge from submit-queue
Use indirect streaming path for remote CRI shim
Last step for https://github.com/kubernetes/kubernetes/issues/29579
- Wire through the remote indirect streaming methods in the docker remote shim
- Add the docker streaming server as a handler at `<node>:10250/cri/{exec,attach,portforward}`
- Disable legacy streaming for dockershim
Note: This requires PR https://github.com/kubernetes/kubernetes/pull/34987 to work.
Tested manually on an E2E cluster.
/cc @euank @feiskyer @kubernetes/sig-node
Automatic merge from submit-queue
Cleanup kubelet eviction manager tests
It cleans up kubelet eviction manager tests
Extracted parts of tests that were similar to each other to functions
No semantic changes, but a lot of shuffling of docstrings to make things
more consistent. In particular, standardise on the zeroth-article (i.e.
prefer `// Version` to `// The version`) and ending all docstrings with
periods.
Provides an opt-in flag, --experimental-fail-swap-on (and corresponding
KubeletConfiguration value, ExperimentalFailSwapOn), which is false by default.
Automatic merge from submit-queue
CRI: Add security context for sandbox/container
Part of #29478. This PR
- adds security context for sandbox and fixes#33139
- encaps container security context to `SecurityContext` and adds missing features
- Note that capability is not fully accomplished in this PR because it is under discussion at #33614.
cc/ @yujuhong @yifan-gu @Random-Liu @kubernetes/sig-node
This allows us to interrupt/kill the executed command if it exceeds the
timeout (not implemented by this commit).
Set timeout in Exec probes. HTTPGet and TCPSocket probes respect the
timeout, while Exec probes used to ignore it.
Add e2e test for exec probe with timeout. However, the test is skipped
while the default exec handler doesn't support timeouts.
Automatic merge from submit-queue
kubelet bootstrap: start hostNetwork pods before we have PodCIDR
Network readiness was checked in the pod admission phase, but pods that
fail admission are not retried. Move the check to the pod start phase.
Issue #35409
Issue #35521
Automatic merge from submit-queue
CRI: rearrange kubelet rutnime initialization
Consolidate the code used by docker+cri and remote+cri for consistency, and to
prevent changing one without the other. Enforce that
`--experimental-runtime-integration-type` has to be set in order for kubelet
use the CRI interface, *even for out-of-process shims`. This simplifies the
temporary `if` logic in kubelet while CRI still co-exists with older logic.
Automatic merge from submit-queue
CRI: Add Status into CRI.
For https://github.com/kubernetes/kubernetes/issues/35701.
Fixes https://github.com/kubernetes/kubernetes/issues/35701.
This PR added a `Status` call in CRI, and the `RuntimeStatus` is defined as following:
``` protobuf
message RuntimeCondition {
// Type of runtime condition.
optional string type = 1;
// Status of the condition, one of true/false.
optional bool status = 2;
// Brief reason for the condition's last transition.
optional string reason = 3;
// Human readable message indicating details about last transition.
optional string message = 4;
}
message RuntimeStatus {
// Conditions is an array of current observed runtime conditions.
repeated RuntimeCondition conditions = 1;
}
```
Currently, only `conditions` is included in `RuntimeStatus`, and the definition is almost the same with `NodeCondition` and `PodCondition` in K8s api.
@yujuhong @feiskyer @bprashanth If this makes sense, I'll send a follow up PR to let dockershim return `RuntimeStatus` and let kubelet make use of it.
@yifan-gu @euank Does this make sense to rkt?
/cc @kubernetes/sig-node
Automatic merge from submit-queue
We only report diskpressure to users, and no longer report inodepressure
See #36180 for more information on why #33218 was reverted.
Automatic merge from submit-queue
CRI: stop sandbox before removing it
Stopping a sandbox includes reclaiming the network resources. By always
stopping the sandbox before removing it, we reduce the possibility of leaking
resources in some corner cases.
Kubernetes Submit Queue