I'm running a fairly resource-intensive service on a Kubernetes cluster to support CI activities. Only a single replica is needed, but it uses a lot of resources (16 cpu), and it's only needed during work hours generally (weekdays, 8am-6pm roughly). My cluster runs in a cloud and is setup with instance autoscaling, so if this service is scaled to zero, that instance can be terminated.
The service is third-party code that cannot be modified (well, not easily). It's a fairly typical HTTP service other than that its work is fairly CPU intensive.
What options exist to automatically scale this Deployment down to zero when idle?
I'd rather not setup a schedule to scale it up/down during working hours because occasionally CI activities are performed outside of the normal hours. I'd like the scaling to be dynamic (for example, scale to zero when idle for >30 minutes, or scale to one when an incoming connection arrives).
Actually Kubernetes supports the scaling to zero only by means of an API call, since the Horizontal Pod Autoscaler does support scaling down to 1 replica only.
Anyway there are a few Operator which allow you to overtake that limitation by intercepting the requests coming to your pods or by inspecting some metrics.
You can take a look at Knative or Keda.
They enable your application to be serverless and they do so in different ways.
Knative, by means of Istio intercept the requests and if there's an active pod serving them, it redirects the incoming request to that one, otherwise it trigger a scaling.
By contrast, Keda best fits event-driven architecture, because it is able to inspect predefined metrics, such as lag, queue lenght or custom metrics (collected from Prometheus, for example) and trigger the scaling.
Both support scale to zero in case predefined conditions are met in a equally predefined window.
Hope it helped.
I ended up implementing a custom solution: https://github.com/greenkeytech/zero-pod-autoscaler
Compared to Knative, it's more of a "toy" project, fairly small, and has no dependency on istio. It's been working well for my use case, though I do not recommend others use it without being willing to adopt the code as your own.
There are a few ways this can be achieved, possibly the most "native" way is using Knative with Istio. Kubernetes by default allows you to scale to zero, however you need something that can broker the scale-up events based on an "input event", essentially something that supports an event driven architecture.
You can take a look at the offcial documents here: https://knative.dev/docs/serving/configuring-autoscaling/
The horizontal pod autoscaler currently doesn’t allow setting the minReplicas field to 0, so the autoscaler will never scale down to zero, even if the pods aren’t doing anything. Allowing the number of pods to be scaled down to zero can dramatically increase the utilization of your hardware.
When you run services that get requests only once every few hours or even days, it doesn’t make sense to have them running all the time, eating up resources that could be used by other pods.
But you still want to have those services available immediately when a client request comes in.
This is known as idling and un-idling. It allows pods that provide a certain service to be scaled down to zero. When a new request comes in, the request is blocked until the pod is brought up and then the request is finally forwarded to the pod.
Kubernetes currently doesn’t provide this feature yet, but it will eventually.
based on documentation it does not support minReplicas=0 so far. read this thread :-https://github.com/kubernetes/kubernetes/issues/69687. and to setup HPA properly you can use this formula to setup required pod :-
desiredReplicas = ceil[currentReplicas * ( currentMetricValue / desiredMetricValue )]
you can also setup HPA based on prometheus metrics follow this link:-
https://itnext.io/horizontal-pod-autoscale-with-custom-metrics-8cb13e9d475
Related
I'm using Knative serving with KPA. Autoscaling is available in Knative based on concurrency and RPS. But we need to scale different services based on queue lengths because there are long running async processes. Is there any way we can achieve this in Knative?
I can't use Knative HPA because we need scale to zero feature of Knative.
Thanks in advance!
If you have async (background or scheduled processes), it's likely that Knative is not a good match for your application. There has been some investigation into exposing the HPA v2 custom metrics scanning options (which would might preclude scale to zero, as you note), but even with HPA2 scaling, you'll still run into problems.
The problem with background processes is that Knative and Kubernetes don't have visibility into which Pods are still doing work, so they are equally likely to shut down a Pod doing work as one that is idle.
One workaround would be to move the async work to be synchronous with a request (possibly by using eventing to send a "do work" event), and then processing those events synchronously -- the eventing Broker won't get upset if your requests take a long time to complete. If you're worried about non-uniform processing times, you can even run a second copy of the Knative Service just for handling the long-running requests.
We have a deployment stack with about 20 microservices/pods. Each deployment goes to its own namespace. To make sure that the cpu and memory are guaranteed for each pod and not shared, we set the request amounts the same as limit amount. Now we sometimes need to deploy more stack into the same performance cluster, e.g. testing different releases of the same stack. The question is whether having more than one deployment in one cluster can invalidate the test result due to shared network or some other reasons?
Initially we were thinking to create one cluster for each performance testing to make sure it is isolated and test results are correct but creating a new cluster and maintaining it a very costly. We also thought about making sure each deployment goes to one node to avoid load testing on one stack impact the others but I'm not sure if that really helps. Please share your knowledge on this as Kubernetes is almost new to us.
If the containers are running on the same underlying hosts then bleedthrough is always possible. If you set all pods into Guaranteed QoS mode (aka requests == limits) then it at least reduces the bleedthrough to a minimum. Running things on one cluster is always fine but if you want to truly reduce the crosstalk to zero then you would need dedicated workload nodes for each.
This is a purely theoretical question. A standard Kubernetes clusted is given with autoscaling in place. If memory goes above a certain targetMemUtilizationPercentage than a new pod is started and it takes on the flow of requests that is coming to the contained service. The number of minReplicas is set to 1 and the number of maxReplicas is set to 5.
What happens when the number of pods that are online reaches maximum (5 in our case) and requests from clients are still coming towards the node? Are these requests buffered somewhere of they are discarded? Can I take any actions to avoid request loss?
Natively Kubernetes does not support messaging queue buffering. Depends on the scenario and setup you use your requests will most likely 'timeout'. To efficiently manage those you`ll need custom resource running inside Kubernetes cluster.
In that situations it very common to use a message broker which ensures communication between microservices is reliable and stable, that the messages are managed and monitored within the system and that messages don’t get lost.
RabbitMQ, Kafka and Redis appears to be most popular but choosing the right one will heaving depend on your requirement and features needed.
Worth to note since Kubernetes essentially runs on linux is that linux itself also manages/limits the requests coming in socket. You may want to read more about it here.
Another thing is that if you have pods limits set or lack of resource it is most likely that pods might be restarted or cluster will become unstable. Usually you can prevent it by configuring some kind of "circuit breaker" to limit amount of requests that could go to backed without overloading it. If the amount of requests goes beyond the circuit breaker threshold, excessive requests will be dropped.
It is better to drop some request than having cascading failure.
I managed to test this scenario and I get 503 Service Unavailable and 403 Forbidden on my requests that do not get processed.
Knative Serving actually does exactly this. https://github.com/knative/serving/
It buffers requests and informs autoscaling decisions based on in-flight request counts. It also can enforce per-Pod max in-flight requests and hold onto request until newly scaled-up Pods come up and then Knative proxies the request to them as it has this container named queue-proxy as a sidecar to its workload type called "Service".
I have what I believe is a simple goal, but I can't figure out how to get Kubernetes to play ball.
For my particular application, I am trying to deploy a number of replicas of a docker image that is a worker for another service. This system uses the hostname of the worker to distinguish between workers that are running at the same time.
I would like to be able to deploy a cluster where every node runs a worker for this service.
The problem is that the master also keeps track of every worker that ever worked for it, and displays these in a status dashboard. The intent is that you spin up a fixed number of workers by hand and leave it that way. I would like to be able to resize my cluster and have the number of workers change accordingly.
This seems like a perfect application for DaemonSet, except that then the hostnames are randomly generated and the master ends up tracking many orphaned hostnames.
An alternative might be StatefulSet, which gives us deterministic hostnames, but I can't find a way to force it to scale to one pod per node.
The system I am running is open source and I am looking into changing how it identifies workers to avoid this mess, but I was wondering if there was any sensible way to dynamically scale a StatefulSet to the number of nodes in the cluster. Or any way to achieve similar functionality.
The one way is to use nodeSelector, but I totally agree with #Markus: the more correct and advanced way is to use anti-affinity. This is really powerful and at the same time simple solution to prevent scheduling pods with the same labels to 1 node.
We have a service which is fairly idle most of the time, hence it would be great for us if we could delete all the pods when the service is not getting any request for say 30 minutes, and in the next time when a new request comes Kubernetes will create the first pod and process the response.
Is it possible to set the min pod instance count to 0?
I found that currently, Kubernetes does not support this, is there a way I can achieve this?
This is not supported in Kubernetes the way it's supported by web servers like nginx, apache or app engines like puma, passenger, gunicorn, unicorn or even Google App Engine Standard where they can be soft started and then brought up the moment the first request comes in with downside of this is that your first requests will always be slower. (There may have been some rationale behind Kubernetes pods not having to behave this way, and I can see a lot of design changes or having to create a new type of workload for this very specific case)
If a pod is sitting idle it would not be consuming that many resources. You could tweak the values of your pod resources for request/limit so that you request a small number of CPUs/Memory and you set the limit to a higher number of CPUs/Memory. The upside of having a pod always running is that in theory, your first requests will never have to wait a long time to get a response.
Yes. You can achieve that using Horizontal Pod Autoscale.
See example of Horizontal Pod Autoscale: Horizontal Pod Autoscaler Walkthrough