Osiris - A general purpose, Scale to Zero component for Kubernetes
Osiris enables greater resource efficiency within a Kubernetes cluster by
allowing idling workloads to automatically scale-to-zero and allowing
scaled-to-zero workloads to be automatically re-activated on-demand by inbound
requests.
Osiris, as a concept, is highly experimental and currently remains under heavy
development.
How it works
Various types of Kubernetes resources can be Osiris-enabled using an annotation.
Osiris-enabled pods are automatically instrumented with a metrics-collecting
proxy deployed as a sidecar container.
Osiris-enabled deployments (if already scaled to a configurable minimum number
of replicas-- one by default) automatically have metrics from their pods
continuously scraped and analyzed by the zeroscaler component. When the
aggregated metrics reveal that all of the deployment's pods are idling, the
zeroscaler scales the deployment to zero replicas.
Under normal circumstances, scaling a deployment to zero replicas poses a
problem: any services that select pods from that deployment (and only that
deployment) would lose all of their endpoints and become permanently
unavailable. Osiris-enabled services, however, have their endpoints managed by
the Osiris endpoints controller (instead of Kubernetes' built-in endpoints
controller). The Osiris endpoints controller will automatically add Osiris
activator endpoints to any Osiris-enabled service that has lost the rest of
its endpoints.
The Osiris activator component receives traffic for Osiris-enabled services
that are lacking any application endpoints. The activator initiates a scale-up
of a corresponding deployment to a configurable minimum number of replicas (one,
by default). When at least one application pod becomes ready, the request will
be forwarded to the pod.
After the activator "reactivates" the deployment, the endpoints controller
(described above) will naturally observe the availability of application
endpoints for any Osiris-enabled services that select those pods and will
remove activator endpoints from that service. All subsequent traffic for the
service will, once again, flow directly to application pods... until a period of
inactivity causes the zeroscaler to take the application offline again.
Scaling to zero and the HPA
Osiris is designed to work alongside the Horizontal Pod Autoscaler and
is not meant to replace it-- it will scale your pods from n to 0 and from 0 to
n, where n is a configurable minimum number of replicas (one, by default). All
other scaling decisions may be delegated to an HPA, if desired.
This diagram better illustrates the different roles of Osiris, the HPA and the
Cluster Autoscaler:
Setup
Prerequisites:
- Helm (v3.0.0 or greater)
- A running Kubernetes cluster.
Installation
Osiris' Helm chart is hosted in an
Azure Container Registry,
which does not yet support anonymous access to charts therein. Until this is
resolved, adding the Helm repository from which Osiris can be installed requires
use of a shared set of read-only credentials.
Make sure helm is initialized in your running kubernetes cluster.
For more details on initializing helm, Go here
helm repo add osiris https://osiris.azurecr.io/helm/v1/repo \
--username eae9749a-fccf-4a24-ac0d-6506fe2a6ab3 \
--password =s-e.2-84BhIo6LM6=/l4C_sFzxb=sT[
Installation requires use of the --devel
flag to indicate pre-release versions
of the specified chart are eligible for download and installation. The following
command install the latest version of Osiris with the default values for all
options - see the next section for all available installation options.
kubectl create namespace osiris-system
helm install osiris osiris/osiris-edge \
--namespace osiris-system \
--devel
Installation Options
Osiris global configuration is minimal - because most of it will be done by the users
with annotations on the Kubernetes resources.
The following table lists the configurable parameters of the Helm chart and their default values.
Parameter |
Description |
Default |
zeroscaler.metricsCheckInterval |
The interval in which the zeroScaler would repeatedly track the pod http request metrics. The value is the number of seconds of the interval. Note that this can also be set on a per-deployment basis, with an annotation. |
150 |
Example of installation with Helm and a custom configuration:
kubectl create namespace osiris-system
helm install osiris osiris/osiris-edge \
--namespace osiris-system \
--devel \
--set zeroscaler.metricsCheckInterval=600
Usage
Osiris will not affect the normal behavior of any Kubernetes resource without
explicitly being directed to do so.
To enabled the zeroscaler to scale a deployment with idling pods to zero
replicas, annotate the deployment like so:
apiVersion: apps/v1
kind: Deployment
metadata:
namespace: my-aoo
name: my-app
annotations:
osiris.deislabs.io/enabled: "true"
spec:
replicas: 1
selector:
matchLabels:
app: my-app
template:
metadata:
labels:
app: nginx
annotations:
osiris.deislabs.io/enabled: "true"
# ...
# ...
Note that the template for the pod also uses an annotation to enable Osiris--
in this case, it enables the metrics-collecting proxy sidecar container on all
of the deployment's pods.
In Kubernetes, there is no direct relationship between deployments and services.
Deployments manage pods and services may select pods managed by one or more
deployments. Rather than attempt to infer relationships between deployments and
services and potentially impact service behavior without explicit consent,
Osiris requires services to explicitly opt-in to management by the Osiris
endpoints controller. Such services must also utilize an annotation to indicate
which deployment should be reactivated when the activator component intercepts a
request on their behalf. For example:
kind: Service
apiVersion: v1
metadata:
namespace: my-namespace
name: my-app
annotations:
osiris.deislabs.io/enabled: "true"
osiris.deislabs.io/deployment: my-app
spec:
selector:
app: my-app
# ...
Configuration
Most of Osiris configuration is done with Kubernetes annotations - as seen in the Usage section.
Deployment Annotations
The following table lists the supported annotations for Kubernetes Deployments
and their default values.
Annotation |
Description |
Default |
osiris.deislabs.io/enabled |
Enable the zeroscaler component to scrape and analyze metrics from the deployment's pods and scale the deployment to zero when idle. Allowed values: y , yes , true , on , 1 . |
no value (= disabled) |
osiris.deislabs.io/minReplicas |
The minimum number of replicas to set on the deployment when Osiris will scale up. If you set 2 , Osiris will scale the deployment from 0 to 2 replicas directly. Osiris won't collect metrics from deployments which have more than minReplicas replicas - to avoid useless collections of metrics. |
1 |
osiris.deislabs.io/metricsCheckInterval |
The interval in which Osiris would repeatedly track the pod http request metrics. The value is the number of seconds of the interval. Note that this value override the global value defined by the zeroscaler.metricsCheckInterval Helm value. |
value of the zeroscaler.metricsCheckInterval Helm value |
Pod Annotations
The following table lists the supported annotations for Kubernetes Pods
and their default values.
Annotation |
Description |
Default |
osiris.deislabs.io/enabled |
Enable the metrics collecting proxy sidecar container to be injected into this pod. Allowed values: y , yes , true , on , 1 . |
no value (= disabled) |
osiris.deislabs.io/ignoredPaths |
The list of (url) paths that should be "ignored" by Osiris. Requests to such paths won't be "counted" by the proxy. Format: comma-separated string. |
no value |
Service Annotations
The following table lists the supported annotations for Kubernetes Services
and their default values.
Annotation |
Description |
Default |
osiris.deislabs.io/enabled |
Enable this service's endpoints to be managed by the Osiris endpoints controller. Allowed values: y , yes , true , on , 1 . |
no value (= disabled) |
osiris.deislabs.io/deployment |
Name of the deployment which is behind this service. This is required to map the service with its deployment. |
no value |
osiris.deislabs.io/loadBalancerHostname |
Map requests coming from a specific hostname to this service. Note that if you have multiple hostnames, you can set them with different annotations, using osiris.deislabs.io/loadBalancerHostname-1 , osiris.deislabs.io/loadBalancerHostname-2 , ... |
no value |
osiris.deislabs.io/ingressHostname |
Map requests coming from a specific hostname to this service. If you use an ingress in front of your service, this is required to create a link between the ingress and the service. Note that if you have multiple hostnames, you can set them with different annotations, using osiris.deislabs.io/ingressHostname-1 , osiris.deislabs.io/ingressHostname-2 , ... |
no value |
osiris.deislabs.io/ingressDefaultPort |
Custom service port when the request comes from an ingress. Default behaviour if there are more than 1 port on the service, is to look for a port named http , and fallback to the port 80 . Set this if you have multiple ports and using a non-standard port with a non-standard name. |
no value |
osiris.deislabs.io/tlsPort |
Custom port for TLS-secured requests. Default behaviour if there are more than 1 port on the service, is to look for a port named https , and fallback to the port 443 . Set this if you have multiple ports and using a non-standard TLS port with a non-standard name. |
no value |
Note that you might see an osiris.deislabs.io/selector
annotation - this is for internal use only, and you shouldn't try to set/update or delete it.
Demo
Deploy the example application hello-osiris
:
kubectl create -f ./example/hello-osiris.yaml
This will create an Osiris-enabled deployment and service named hello-osiris
.
Get the External IP of the hello-osiris
service once it appears:
kubectl get service hello-osiris -o jsonpath='{.status.loadBalancer.ingress[*].ip}'
Point your browser to "http://<EXTERNAL-IP>"
, and verify that
hello-osiris
is serving traffic.
After about 2.5 minutes, the Osiris-enabled deployment should scale to zero
replicas and the one hello-osiris
pod should be terminated.
Make a request again, and watch as Osiris scales the deployment back to one
replica and your request is handled successfully.
Limitations
It is a specific goal of Osiris to enable greater resource efficiency within
Kubernetes clusters, in general, but especially with respect to "nodeless"
Kubernetes options such as
Virtual Kubelet or
Azure Kubernetes Service Virtual Nodes, however,
due to known issues with those technologies, Osiris remains incompatible with
them for the near term.
Contributing
Osiris follows the CNCF Code of Conduct.