Change the NFD API handler to re-try on node update failures. Will work
around transient failures, making sure that failed nodes (i.e. nodes
that we failed to update) don't need to wait for the 1 hour resync
period before being tried again.
Increase the NFD API controller resync period from 5 minutes to 1 hour.
The resync causes nfd-master to replay all NodeFeature and
NodeFeatureRule objects, being effectively a "big hammer reset all"
button. This should only be needed as an "insurance" to fix labels et al
in case they have been manually tampered (outside NFD) and against
certain bugs in nfd itself. NFD is not supposed to manage anything
fast-changing so 1 hour should be enough.
This change only affects behavior when the NodeFeature API has been
enabled (with -enable-nodefeature-api).
Add support for management of Extended Resources via the
NodeFeatureRule CRD API.
There are usage scenarios where users want to advertise features
as extended resources instead of labels (or annotations).
This patch enables the discovery of extended resources, via annotation
and patch of node.status.capacity and node.status.allocatable. By using
the NodeFeatureRule API.
Co-authored-by: Carlos Eduardo Arango Gutierrez <eduardoa@nvidia.com>
Co-authored-by: Markus Lehtonen <markus.lehtonen@intel.com>
Co-authored-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Signed-off-by: Carlos Eduardo Arango Gutierrez <eduardoa@nvidia.com>
Update node status before node metadata. This fixes a problem where we
lose track of NFD-managed extended resources in case patching node
status fails. Previously we removed all labels and annotations
(including the one listing our ERs) and only after that updated node
status. If node status update failed we had lost the annotation but
extended resources were still there, leaving them orphaned.
Disallow taints having a key with "kubernetes.io/" or "*.kubernetes.io/"
prefix. This is a precaution to protect the user from messing up with
the "official" well-known taints from Kubernetes itself. The only
exception is that the "nfd.node.kubernetes.io/" prefix is allowed.
However, there is one allowed NFD-specific namespace (and its
sub-namespaces) i.e. "feature.node.kubernetes.io" under the
kubernetes.io domain that can be used for NFD-managed taints.
Also disallow unprefixed taint keys. We don't add a default prefix to
unprefixed taints (like we do for labels) from NodeFeatureRules. This is
to prevent unpleasant surprises to users that need to manage matching
tolerations for their workloads.
Similar to the nfd-worker, in this PR we want to support the
dynamic run-time configurability through a config file for the nfd-master.
We'll use a json or yaml configuration file along with the fsnotify in
order to watch for changes in the config file. As a result, we're
allowing dynamic control of logging params, allowed namespaces,
extended resources, label whitelisting, and denied namespaces.
Signed-off-by: AhmedGrati <ahmedgrati1999@gmail.com>
Correctly handle the case where no NodeFeature objects exist for certain
node (and NodeFeature API has been enabled with
-enable-nodefeature-api). In this case all the labels should be removed.
We want to always update all nodes at startup. Without this patch we
don't get any update event from the controller if no NodeFeature or
NodeFeatureRule objects exist in the cluster. Thus all nodes would stay
untouched whereas we really want to remove all labels from all nodes in
this case.
Implement a naive ratelimiter for node update events originating from
the nfd API. We might get a ton of events in short interval. The
simplest example is startup when we get a separate Add event for every
NodeFeature and NodeFeatureRule object. Without rate limiting we
run "update all nodes" separately for each NodeFeatureRule object, plus,
we would run "update node X" separately for each NodeFeature object
targeting node X. This is a huge amount of wasted work because in
principle just running "update all nodes" once should be enough.
Implement handling of multiple NodeFeature objects by merging all
objects (targeting a certain node) into one before processing the data.
This patch implements MergeInto() methods for all required data types.
With support for multiple NodeFeature objects per node, The "nfd api
workflow" can be easily demonstrated and tested from the command line.
Creating the folloiwing object (assuming node-n exists in the cluster):
apiVersion: nfd.k8s-sigs.io/v1alpha1
kind: NodeFeature
metadata:
labels:
nfd.node.kubernetes.io/node-name: node-n
name: my-features-for-node-n
spec:
# Features for NodeFeatureRule matching
features:
flags:
vendor.domain-a:
elements:
feature-x: {}
attributes:
vendor.domain-b:
elements:
feature-y: "foo"
feature-z: "123"
instances:
vendor.domain-c:
elements:
- attributes:
name: "elem-1"
vendor: "acme"
- attributes:
name: "elem-2"
vendor: "acme"
# Labels to be created
labels:
vendor-feature.enabled: "true"
vendor-setting.value: "100"
will create two feature labes:
feature.node.kubernetes.io/vendor-feature.enabled: "true"
feature.node.kubernetes.io/vendor-setting.value: "100"
In addition it will advertise hidden/raw features that can be used for
custom rules in NodeFeatureRule objects. Now, creating a NodeFeatureRule
object:
apiVersion: nfd.k8s-sigs.io/v1alpha1
kind: NodeFeatureRule
metadata:
name: my-rule
spec:
rules:
- name: "my feature rule"
labels:
"my-feature": "true"
matchFeatures:
- feature: vendor.domain-a
matchExpressions:
feature-x: {op: Exists}
- feature: vendor.domain-c
matchExpressions:
vendor: {op: In, value: ["acme"]}
will match the features in the NodeFeature object above and cause one
more label to be created:
feature.node.kubernetes.io/my-feature: "true"
Deprecate the '-featurerules-controller' command line flag as the name
does not describe the functionality anymore: in practice it controls the
CRD controller handling both NodeFeature and NodeFeatureRule objects.
The patch introduces a duplicate, more generally named, flag
'-crd-controller'. A warning is printed in the log if
'-featurerules-controller' flag is encountered.
Add initial support for handling NodeFeature objects. With this patch
nfd-master watches NodeFeature objects in all namespaces and reacts to
changes in any of these. The node which a certain NodeFeature object
affects is determined by the "nfd.node.kubernetes.io/node-name"
annotation of the object. When a NodeFeature object targeting certain
node is changed, nfd-master needs to process all other objects targeting
the same node, too, because there may be dependencies between them.
Add a new command line flag for selecting between gRPC and NodeFeature
CRD API as the source of feature requests. Enabling NodeFeature API
disables the gRPC interface.
-enable-nodefeature-api enable NodeFeature CRD API for incoming
feature requests, will disable the gRPC
interface (defaults to false)
It is not possible to serve gRPC and watch NodeFeature objects at the
same time. This is deliberate to avoid labeling races e.g. by nfd-worker
sending gRPC requests but NodeFeature objects in the cluster
"overriding" those changes (labels from the gRPC requests will get
overridden when NodeFeature objects are processed).
Drop the gRPC communication to nfd-master and connect to the Kubernetes
API server directly when updating NodeResourceTopology objects.
Topology-updater already has connection to the API server for listing
Pods so this is not that dramatic change. It also simplifies the code
a lot as there is no need for the NFD gRPC client and no need for
managing TLS certs/keys.
This change aligns nfd-topology-updater with the future direction of
nfd-worker where the gRPC API is being dropped and replaced by a
CRD-based API.
This patch also update deployment files and documentation to reflect
this change.
This commits extends NFD master code to support adding node taints
from NodeFeatureRule CR. We also introduce a new annotation for
taints which helps to identify if the taint set on node is owned
by NFD or not. When user deletes the taint entry from
NodeFeatureRule CR, NFD will remove the taint from the node. But
to avoid accidental deletion of taints not owned by the NFD, it
needs to know the owner. Keeping track of NFD set taints in the
annotation can be used during the filtering of the owner. Also
enable-taints flag is added to allow users opt in/out for node
tainting feature. The flag takes precedence over taints defined
in NodeFeatureRule CR. In other words, if enbale-taints is set to
false(disabled) and user still defines taints on the CR, NFD will
ignore those taints and skip them from setting on the node.
Signed-off-by: Feruzjon Muyassarov <feruzjon.muyassarov@intel.com>
Flatten the data structure that stores features, dropping the "domain"
level from the data model. That extra level of hierarchy brought little
benefit but just caused some extra complexity, instead. The new
structure nicely matches what we have in the NodeFeatureRule object (the
matchFeatures field of uses the same flat structure with the "feature"
field having a value <domain>.<feature>, e.g. "kernel.version").
This is pre-work for introducing a new "node feature" CRD that contains
the raw feature data. It makes the life of both users and developers
easier when both CRDs, plus our internal code, handle feature data in a
similar flat structure.
Move the previously-protobuf-only internal "feature api" over to the
public "nfd api" package. This is in preparation for introducing a new
CRD API for communicating features.
This patch carries no functional changes. Just moving code around.
Refactor the code so that the initialization and running of the gRPC
server is done in a separate function. The goal is to make the code more
maintainable in terms of disabling (and eventually removing) the gRPC
functionality in the future.
Remove the cleanup code that removes ancient NFD labels with the
node.alpha.kubernetes-incubator.io/ prefix. This label namespace was
deprecated/dropped already in v0.4.0 so it should be safe to drop this
code.
Use the single-dash (i.e. '-option' instead of '--option') format
consistently accross log messages and documentation. This is the format
that was mostly used, already, and shown by command line help of the
binaries, for example.
Support backreferencing of output values from previous rules. Enables
complex rule setups where custom features are further combined together
to form even more sophisticated higher level labels. The labels created
by preceding rules are available as a special 'rule.matched' feature
(for matchFeatures to use).
If referencing rules accross multiple configs/CRDs care must be taken
with the ordering. Processing order of rules in nfd-worker:
1. Static rules
2. Files from /etc/kubernetes/node-feature-discovery/custom.d/
in alphabetical order. Subdirectories are processed by reading their
files in alphabetical order.
3. Custom rules from main nfd-worker.conf
In nfd-master, NodeFeatureRule objects are processed in alphabetical
order (based on their metadata.name).
This patch also adds new 'vars' fields to the rule spec. Like 'labels',
it is a map of key-value pairs but no labels are generated from these.
The values specified in 'vars' are only added for backreferencing into
the 'rules.matched' feature. This may by desired in schemes where the
output of certain rules is only used as intermediate variables for other
rules and no labels out of these are wanted.
An example setup:
- name: "kernel feature"
labels:
kernel-feature:
matchFeatures:
- feature: kernel.version
matchExpressions:
major: {op: Gt, value: ["4"]}
- name: "intermediate var feature"
vars:
nolabel-feature: "true"
matchFeatures:
- feature: cpu.cpuid
matchExpressions:
AVX512F: {op: Exists}
- feature: pci.device
matchExpressions:
vendor: {op: In, value: ["8086"]}
device: {op: In, value: ["1234", "1235"]}
- name: top-level-feature
matchFeatures:
- feature: rule.matched
matchExpressions:
kernel-feature: "true"
nolabel-feature: "true"
Enable Custom Resource based label creation in nfd-master. This extends
the previously implemented controller stub for watching NodeFeatureRule
objects. NFD-master watches NodeFeatureRule objects in the cluster and
processes the rules on every incoming labeling request from workers.
The functionality relies on the "raw features" (identical to how
nfd-worker handles custom rules) submitted by nfd-worker, making it
independent of the label source configuration of the worker. This means
that the labeling functions as expected even if all sources in the
worker are disabled.
NOTE: nfd-master is stateless and re-labeling only happens on the
reception of SetLabelsRequest from the worker – i.e. on intervals
specified by the core.sleepInterval configuration option (or
-sleep-interval cmdline flag) of each nfd-worker instance. This means
that modification/creation of NodeFeatureRule objects does not
automatically update the node labels. Instead, the changes only come
visible when workers send their labeling requests.
Add a new command line flag for disabling/enabling the controller for
NodeFeatureRule objects. In practice, disabling the controller disables
all labels generated from rules in NodeFeatureRule objects.
Implement a simple controller stub that operates on NodeFeatureRule
objects. The controller does not yet have any functionality other than
logging changes in the (NodeFeatureRule) objecs it is watching.
Also update the documentation on the -no-publish flag to match the new
functionality.
The NodeResourceTopology API has been made cluster
scoped as in the current context a CR corresponds to
a Node and since Node is a cluster scoped resource it
makes sense to make NRT cluster scoped as well.
Ref: https://github.com/k8stopologyawareschedwg/noderesourcetopology-api/pull/18
Signed-off-by: Swati Sehgal <swsehgal@redhat.com>
There have been recent changes made to the noderesourcetopology API
storing the proto file generated using go-to-protobuf tool and
this code inports the proto generated in the API in the topology-updater.proto
The PRs corresponding to the changes are as follows:
https://github.com/k8stopologyawareschedwg/noderesourcetopology-api/pull/9https://github.com/k8stopologyawareschedwg/noderesourcetopology-api/pull/13
Commands used to generate topology-updater.pb.go file:
go install github.com/golang/protobuf/protoc-gen-go@v1.4.3
go mod vendor
protoc --go_opt=paths=source_relative --go_out=plugins=grpc:. pkg/topologyupdater/topology-updater.proto -I. -Ivendor
As part of implmentation of this patch, reserved (non-allocatable) CPUs
are evaluated by performing a difference between all the CPUs on a system
(determined by using ghw) and allocatable CPUs (determined by querying
GetAllocatableResources podResource API endpoint).
When aggregator creates the NUMA zones, it will skip the zone creation if
there are no allocatable resources. In this update we creates those missing
zone with zero allocatable/available resources so we won't have holes in the
array of reported zones.
Co-Authored-by: Talor Itzhak <titzhak@redhat.com>
Signed-off-by: Swati Sehgal <swsehgal@redhat.com>
- This patch allows to expose Resource Hardware Topology information
through CRDs in Node Feature Discovery.
- In order to do this we introduce another software component called
nfd-topology-updater in addition to the already existing software
components nfd-master and nfd-worker.
- nfd-master was enhanced to communicate with nfd-topology-updater
over gRPC followed by creation of CRs corresponding to the nodes
in the cluster exposing resource hardware topology information
of that node.
- Pin kubernetes dependency to one that include pod resource implementation
- This code is responsible for obtaining hardware information from the system
as well as pod resource information from the Pod Resource API in order to
determine the allocatable resource information for each NUMA zone. This
information along with Costs for NUMA zones (obtained by reading NUMA distances)
is gathered by nfd-topology-updater running on all the nodes
of the cluster and propagate NUMA zone costs to master in order to populate
that information in the CRs corresponding to the nodes.
- We use GHW facilities for obtaining system information like CPUs, topology,
NUMA distances etc.
- This also includes updates made to Makefile and Dockerfile and Manifests for
deploying nfd-topology-updater.
- This patch includes unit tests
- As part of the Topology Aware Scheduling work, this patch captures
the configured Topology manager scope in addition to the Topology manager policy.
Based on the value of both attribues a single string will be populated to the CRD.
The string value will be on of the following {SingleNUMANodeContainerLevel,
SingleNUMANodePodLevel, BestEffort, Restricted, None}
Co-Authored-by: Artyom Lukianov <alukiano@redhat.com>
Co-Authored-by: Francesco Romani <fromani@redhat.com>
Co-Authored-by: Talor Itzhak <titzhak@redhat.com>
Signed-off-by: Swati Sehgal <swsehgal@redhat.com>
Add a separate label namespace for profile labels, intended for
user-specified higher level "meta features". Also sub-namespaces of this
(i.e. <sub-ns>.profile.node.kubernetes.io) are allowed.
Allow <sub-ns>.feature.node.kubernetes.io label namespaces. Makes it
possible to have e.g. vendor specific label ns without the need to user
-extra-label-ns.
This can be used to help running multiple parallel NFD deployments in
the same cluster. The flag changes the node annotation namespace to
<instance>.nfd.node.kubernetes.io allowing different nfd-master intances
to store metadata in separate annotations.
Handle both creation and parsing of the "feature-labels" and
"extended-resources" annotations in the function. I think this is more
logical to keep them together.
When updating node labels and annotations use JSON patches instead of
doing a read-modify-write on the whole node object. Patching is already
being used in managing extended resources so some of the existing code
was re-usable.
This patch should mitigate the problem of node update failures caused by
race conditions (a change in the node object between our read and write)
resulting e.g. in errors/restarts in nfd worker pods.
For historical reasons the labels in the default nfd namespace have been
internally represented without the namespace part. I.e. instead of
"feature.node.kubernetes.io/foo" we just use "foo". NFD worker uses this
representation, too, both internally and over the gRPC requests. The
same scheme has been used for annotations.
This patch changes NFD master to use fully namespaced label and
annotation names internally. This hopefully makes the code a bit more
understandable. It also addresses some corner cases making the handling
of label names consistent, making it possible to use both "truncated"
and fully namespaced names over the gRPC interface (and in the
annotations).
A new sub-command like flag for cleaning up a cluster. When --prune is
specified nfd-master removes all NFD related labels, annotations and
extended resources from all nodes of the cluster and exits.
This should help undeployment of NFD and be useful for development.
This adds support for making selected labels extended resources.
Labels which have integer values, can be promoted to Kubernetes extended
resources by listing them to the added command line flag
`--resource-labels`. These labels won't then show in the node label
section, they will appear only as extended resources.
Signed-off-by: Ukri Niemimuukko <ukri.niemimuukko@intel.com>
Also, adds new method WaitForReady() into NfdMaster.
In practice, this quite widely tests nfd-master, too, as the tests
create an instance of NfdMaster and verify that the communication
between master and worker works.
Move most of the code under cmd/nfd-master and cmd/nfd-worker into new
packages pkg/nfd-master and pk/nfd-worker, respectively. Makes extending
unit tests to "main" functions easier.
