Add a way to set resource limits/requests on running pods
Ref: https://github.com/kubernetes/kubernetes/issues/21648
I squashed the commits to make rebasing easier
Change log:
- fixed a typo that caused the command to be run with kubectl set set instead of the correct kubectl set limit
- added a ResourcesWithPodTemplates to pkg/kubectl/cmd/util/factory.go
instead of hardcoding these resources move there description all in one place
- Fixing some of the flow control in kubectl set limit
- update the help info
- changed the name of ResourcesWithPodTemplates to ResourcesWithPodSpecs to more accuratly describe what it is doing
and changed the variable names to lower case to conform to go's variable naming convention
- changing the name of the command from 'set limit' to 'set resources'
- Adding the new file pkg/kubectl/cmd/set/set_resources.go
- changes to the test cases to reflect the change from 'kubectl set limit' to 'kubectl set resources'
- comment removed
- adding the man page to the git repository attempting to fix Jenkins tests
- adding the user guide
- fixed a few typos
- typo in hack/cmd-test.sh
- implamenting suggestions for command help text
- adding the dry-run flag
- removing the "remove" option in favor of zeroing out request/limits in order to remove them
- changed limits/requests to requests/limit
- changing ResourcesWithPodSpec
- updated generated docs and removed whitespace
- change priint on success message from "resource limits/requests updated" to "resource requirements updated"
- minor rebasing issues - 'hack/test-cmd.sh' now passes
- cmdutil.PrintSuccess added another argument
- fixing mungedocs failure
- removed whitespace from hack/make-rules/test-cmd.sh and an erroneous entry from pkg/cloudprovider/providers/openstack/MAINTAINERS.md
- fixed typo in Short: field of the cobra command
- rebased
- Creating a new factory in the ResourcesWithPodSpecs() so that the testing will pass
- changing ResourcesWithPodSpecs, it doesn't need to be a method of factory
Automatic merge from submit-queue
Log more information on pod status updates
Also bump the logging level to V2 so that we can see them in a non-test
cluster.
Automatic merge from submit-queue
azure: lower log priority for skipped nic update message
**What this PR does / why we need it**: Very minor, just wanted to remove some log noise I introduced in #34526.
I chose `V(3)` since it aligns with the other nicupdate message printed out here, and will be hidden for the usual default of `--v=2`.
**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
```
Automatic merge from submit-queue
Merge string flag into util flag
Continuing my work on https://github.com/kubernetes/kubernetes/issues/15634
This refactoring is expected to be completely finished and then I will add a verify scripts in `hack`
Automatic merge from submit-queue
split genericapiserver configuration apart so that you can run withou…
…t flag options
@dims Take a look at this re-slicing of the `genericapiserver.Config` creation. I think this helps composers overall and resolves the chicken/egg problem you were having.
Automatic merge from submit-queue
Assign options.Err in "set image"
**What this PR does / why we need it**:
There is a usage of options.Err in a Printf, but this option is never set.
This patch passes the stderr into the command and assigns the option correctly.
**Which issue this PR fixes**
fixes#34433
**Special notes for your reviewer**:
None
**Release note**:
```release-note
```
Automatic merge from submit-queue
add UpdateRuntimeConfig interface
Expose UpdateRuntimeConfig interface in RuntimeService for kubelet to pass a set of configurations to runtime. Currently it only takes PodCIDR.
The use case is for kubelet to pass configs to runtime. Kubelet holds some config/information which runtime does not have, such as PodCIDR. I expect some of kubelet configurations will gradually move to runtime, but I believe cases like PodCIDR, which dynamically assigned by k8s master, need to stay for a while.
Automatic merge from submit-queue
Add e2e tests for storageclass
- test pd-ssd and pd-standard on GCE,
- test all four volume types and encryption on AWS
- test just the default volume type on OpenStack (right now, there is no API
to get list of them)
These tests are quite slow, e.g. there are two tests on AWS that has to run mkfs.ext4 on 500 GB magnetic drive with low IOPS, which takes ~3-4 minutes each.
There is a usage of options.Err in a Printf, but this option is never set.
This patch passes the stderr into the command and assigns the option correctly.
Automatic merge from submit-queue
remove call to compinit in zsh completion output
**Release note**:
```release-note
release-note-none
```
Fixes: https://github.com/kubernetes/kubernetes/issues/32029
Fixes:
https://github.com/kubernetes/kubernetes/issues/27538#issuecomment-238574035
The zsh completion output makes a call to "compinit" which causes the
zsh completion system to re-initialize every time `<root_cmd> completion zsh`
is sourced, overwriting any settings already applied to other commands.
This in-turn caused other commands' completions to break (such as git,
gcloud, vim) causing an error "function definition file not found" to
be returned any time a tab-completion was attempted.
This patch removes the call to `compinit` in the zsh completion output,
causing no behavioral changes to the existing `completion` command, but
fixing any issues that were caused after sourcing its output.
This allows security groups to be created and attached to the neutron
port that the loadbalancer is using on the subnet.
The security group ID that is assigned to the nodes needs to be
provided, to allow for traffic from the loadbalancer to the nodePort
to be refelected in the rules.
This adds two config items to the LoadBalancer options -
ManageSecurityGroups (bool)
NodeSecurityGroupID (string)
Automatic merge from submit-queue
kubeadm implement preflight checks
Checks that user running kubeamd init and join is root and will only execute
command if user is root. Moved away from using kubectl error handling to
having kubeadm handle its own errors. This should allow kubeadm to have
more meaningful errors, exit codes, and logging for specific kubeadm use
cases.
fixes#33908
Fixes: https://github.com/kubernetes/kubernetes/issues/32029
Fixes:
https://github.com/kubernetes/kubernetes/issues/27538#issuecomment-238574035
The zsh completion output makes a call to "compinit" which causes the
zsh completion system to re-initialize every time `<root_cmd> completion zsh`
is sourced, overwriting any settings already applied to other commands.
This in-turn caused other commands' completions to break (such as git,
gcloud, vim) causing an error "function definition file not found" to
be returned any time a tab-completion was attempted.
This patch removes the call to `compinit` in the zsh completion output,
causing no behavioral changes to the existing `completion` command, but
fixing any issues that were caused after sourcing its output.
Automatic merge from submit-queue
Change legacy API resource registration
Updates the legacy API resource registration to emphasize its different-ness and to simplify supporting objects. The option has to remain in the genericapiserverconfig for multiple prefixes to enable cases where composers/extenders had composed additional groupless APIs. See OpenShift as an example.
However this is now transparent to "normal" composers.
@ncdc since sttts is out.
Add skip-preflight-checks to known flags.
Fix bug with preflight checks not returning system is-active as errors.
Fix error handling to use correct function.
- test pd-ssd and pd-standard on GCE,
- test all four volume types on AWS
- test just the default volume type on OpenStack (right now, there is no API
to get list of them)
Includes checks for verifying services exist and are enabled, ports are
open, directories do not exist or are empty, and required binaries are
in the path.
Checks that user running kubeamd init and join is root and will only execute
command if user is root. Moved away from using kubectl error handling to
having kubeadm handle its own errors. This should allow kubeadm to have
more meaningful errors, exit codes, and logging for specific kubeadm use
cases.
Automatic merge from submit-queue
Handle DeletedFinalStateUnknown in NodeController
Fix#34692
```release-note
Fix panic in NodeController caused by receiving DeletedFinalStateUnknown object from the cache.
```
cc @davidopp
Automatic merge from submit-queue
WantsAuthorizer admission plugin support
The next step of PSP admission is to be able to limit the PSPs used based on user information. To do this the admission plugin would need to make authz checks for the `user.Info` in the request. This code allows a plugin to request the injection of an authorizer to allow it to make the authz checks.
Note: this could be done with a SAR, however since admission is running in the api server using the SAR would incur an extra hop vs using the authorizer directly.
@deads2k @derekwaynecarr
Automatic merge from submit-queue
Revert "Error out when any RS has more available pods then its spec r…
Reverts https://github.com/kubernetes/kubernetes/pull/29808
The PR is wrong because we can have more available pods than desired every time we scale down.
@kubernetes/deployment ptal
Automatic merge from submit-queue
azure: filter load balancer backend nodes to PrimaryAvailabilitySet (if set)
<!-- Thanks for sending a pull request! Here are some tips for you:
1. If this is your first time, read our contributor guidelines https://github.com/kubernetes/kubernetes/blob/master/CONTRIBUTING.md and developer guide https://github.com/kubernetes/kubernetes/blob/master/docs/devel/development.md
2. If you want *faster* PR reviews, read how: https://github.com/kubernetes/kubernetes/blob/master/docs/devel/faster_reviews.md
3. Follow the instructions for writing a release note: https://github.com/kubernetes/kubernetes/blob/master/docs/devel/pull-requests.md#release-notes
-->
**What this PR does / why we need it**:
- Adds a new field (`PrimaryAvailabilitySetName`) to the Azure CloudProvider config struct
- If the field is set, only machines who are in that availabilitySet are added to the load balancer backend pool.
This is required to:
- Support more than 100 nodes in Azure (only can have 100 nodes per availability set)
- Support multiple availability sets per cluster (An Azure L4 LoadBalancer can only be pointed at nodes in a single availability set)
Without this PR, or if the field is **not** set in a cluster that contains two availabilitysets, then the following is observed:
- Azure resources are created (LB, LB rules, NSG rules, public IP)
- Azure throws errors when trying to add nodes from the "other" availability set
- The service winds up exposed to the outside world (if you manually retrieve the public ip from Azure API)
- Kubernetes controller-manager's service loop keeps retrying forever because it never finishes fully successfully
- The "external ip" property field is never updated.
**Which issue this PR fixes**: Fixes#34293
**Unknowns**:
- Naming convention: `LoadBalancedAvailabilitySet` might be more descriptive than `PrimaryAvailabilitySet`, but is also a misnomer since `kube-proxy` will still end up routing requests to all relevant nodes.
- Is it worth trying to be "smart" about it in the case the user hasn't set this field in the config? Save the first availability set name and try not to add any nodes that aren't also in that one? It may simply be better to just let this fail so the user has to choose the right setting for their use-case.
**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
azure: add PrimaryAvailabilitySet to config, only use nodes in that set in the loadbalancer pool
```
CC: @brendandburns, @anhowe
Automatic merge from submit-queue
return warning on empty list result in kubectl get
**Release note**:
```release-note
NONE
```
The current default behavior of `kubectl get` is to return an empty
output when there are no resources to display. This patch improves
usability by returning a warning through stderr in the case of an empty
list.
##### Before
`$ kubectl get pods`
- *empty output*
##### After
`$ kubectl get pods`
```
There are no resources to display.
```
Automatic merge from submit-queue
Allow kuberuntime to get network namespace for not ready sandboxes
Kubelet calls TearDownPod to clean up the network resources for a pod sandbox.
TearDownPod relies on GetNetNS to retrieve network namespace, and the current
implementation makes this impossible for not-ready sandboxes. This change
removes the unnecessary filter to fix this issue.
Automatic merge from submit-queue
Generate unique Operation IDs for root OpenAPI spec
This PR adds a customization method GetOperationID to OpenAPI spec generation and then use it to make sure root spec has unique operation IDs by mostly adding GroupVersion to the start of operation ID.
Automatic merge from submit-queue
Add PVC storage to LimitRange
This PR adds the ability to add a LimitRange to a namespace that enforces min/max on `pvc.Spec.Resources.Requests["storage"]`.
@derekwaynecarr @abhgupta @kubernetes/sig-storage
Examples forthcoming.
```release-note
pvc.Spec.Resources.Requests min and max can be enforced with a LimitRange of type "PersistentVolumeClaim" in the namespace
```
Automatic merge from submit-queue
kubeadm: fix conversion macros and add kubeadm to round trip testing
Tests are probably broken but I'll fix. @jbeda this probably fixes your change unless we decide we need generated deep copies or conversions.
@kubernetes/sig-cluster-lifecycle
Kubelet calls TearDownPod to clean up the network resources for a pod sandbox.
TearDownPod relies on GetNetNS to retrieve network namespace, and the current
implementation makes this impossible for not-ready sandboxes. This change
removes the unnecessary filter to fix this issue.
