This is a workaround for the issue that the kubelet cannot differentiate
the container statuses of the previous podSandbox from the current one.
If the node is rebooted, all containers will be in the exited state and
the kubelet will try to recreate a new podSandbox. In this case, the
kubelet should not mistakenly think that the newly created podSandbox
has been initialized.
Add retry mechanism to handle cases where after kubelet restarts, the device
plugin unix socket(s) were created but not ready to serve yet.
Signed-off-by: Swati Sehgal <swsehgal@redhat.com>
Add kubeletSocket file to fsnotify instead of polling and waiting for deletion
of device plugin unix socket as a way of detecting kubelet restart. We need to
ensure that the device plugin re-registers itself after kubelet restart depending
on the configured registration mode (auto-registration or controller registration).
Signed-off-by: Swati Sehgal <swsehgal@redhat.com>
If the user specifies the intent to control registration process, we rely on
registration triggers (deletion of control file) to prompt registration.
This behvaiour is expected to be consistent across kubelet restarts and therefore
across the watch calls where we watch for changes to the unix socket so we make
this part of Stub object instead of a parameter.
Co-authored-by: Francesco Romani <fromani@redhat.com>
Signed-off-by: Swati Sehgal <swsehgal@redhat.com>
In case `REGISTER_CONTROL_FILE` is specified, we want to ensure that the
registration is triggered by deletion of the control file. This is
applicable both when the registration happens for the first time and
subsequent ones because of kubelet restarts.
Signed-off-by: Swati Sehgal <swsehgal@redhat.com>
In issue: 115107 we added an environment variable to control the registration of sample
device plugin to kubelet. The intent of this patch is to ensure that the default
behaviour of the plugin is to register to kubelet (in case no environment
variable is specified).
In addition to that, we want to ensure that the plugin registers itself not just once.
It should re-register itself to kubelet in case of node reboot or kubelet restarts.
Signed-off-by: Swati Sehgal <swsehgal@redhat.com>
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>
Include number of requested and available CPUs in the error message
when the assignment of CPUs fails because there are less available
CPUs than requested.
Record volume plugin name when a volume in a Pod needs a different
"mount -o context" value than the actually mounted one.
We expect that NFS, CIFS and CephFS volumes would be able to mount such
volumes just fine with multiple "-o context" values.
We know that the block-volume based ones (ext4, xfs, btrfs, ...) cannot do
that.
Therefore want to distinguish the volume plugin in metrics, anything
block-volume based could break an existing application.
Rate limitter.go file is a generic file implementing
grpc Limiter interface. This file can be reuse by other gRPC
API not only by podresource.
Change-Id: I905a46b5b605fbb175eb9ad6c15019ffdc7f2563
Kubelet, if using cloud provider external, initializes temporary
the node addresses using the non-cloud provider logic, until the
cloud provider overrides it.
This behavior has undesired consequences if the cloud-provider addresses
are different than the original ones, specially for hostNetwork pods,
that inherit these addresses from the Node.
Since some cloud-providers depend on this behavior, in order to keep
backward compatibility, assume that the specifying addresses via
the node-ip flags means that the intent is to keep the existing
behavior to temporary initialize the addresses.
If the node-ips are the unspecified addresses or are not set, then
wait for the external cloud provider to set the node addresses.
Change-Id: I3a3895f9b830769f9658e6a03f058c914c438a09
Signed-off-by: Antonio Ojea <aojea@google.com>
Add file doc.go with some rudimentary information to package
kubelet/cm. This will make it easier for people approaching the
kubelet codebase for the first time to quickly understand what's
in the package, since its name is abbreviated and hostile to
newcomers.
This removes deprecated sets.String and sets.Int
- replace sets.String with sets.Set[string]
- replace sets.Int with sets.Set[int]
- replace sets.NewString with sets.New[string]
- replace sets.NewInt with sets.New[int]
- replace sets.(OLD).List with sets.List(NEW)
This change bypasses all logic to set swap in the linux container
resources if a swap controller is not available on node. Failing
to do so may cause errors in runc when starting a container with
a swap configuration -- even if this is set to 0.
Signed-off-by: Evan Lezar <elezar@nvidia.com>
This change switches to using isCgroup2UnifiedMode locally to ensure
that any mocked function is also used when checking the swap controller
availability.
Signed-off-by: Evan Lezar <elezar@nvidia.com>
