churromorales commented on code in PR #13156: URL: https://github.com/apache/druid/pull/13156#discussion_r995005949
########## docs/development/extensions-contrib/k8s-jobs.md: ########## @@ -0,0 +1,125 @@ +--- +id: k8s-jobs +title: "MM-less Druid in K8s" +--- + +<!-- + ~ Licensed to the Apache Software Foundation (ASF) under one + ~ or more contributor license agreements. See the NOTICE file + ~ distributed with this work for additional information + ~ regarding copyright ownership. The ASF licenses this file + ~ to you under the Apache License, Version 2.0 (the + ~ "License"); you may not use this file except in compliance + ~ with the License. You may obtain a copy of the License at + ~ + ~ http://www.apache.org/licenses/LICENSE-2.0 + ~ + ~ Unless required by applicable law or agreed to in writing, + ~ software distributed under the License is distributed on an + ~ "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + ~ KIND, either express or implied. See the License for the + ~ specific language governing permissions and limitations + ~ under the License. + --> + +Consider this an [EXPERIMENTAL](../experimental.md) feature mostly because it has not been tested yet on a wide variety of long-running Druid clusters. + +Apache Druid Extension to enable using Kubernetes for launching and managing tasks instead of the Middle Managers. This extension allows you to launch tasks as K8s jobs removing the need for your middle manager. + +## How it works + +It takes the podSpec of your `Overlord` pod and creates a kubernetes job from this podSpec. Thus if you have sidecars such as splunk, hubble, istio it can optionally launch a task as a k8s job. All jobs are natively restorable, they are decopled from the druid deployment, thus restarting pods or doing upgrades has no affect on tasks in flight. They will continue to run and when the overlord comes back up it will start tracking them again. + +## Configuration + +To use this extension please make sure to [include](../extensions.md#loading-extensions)`druid-kubernetes-overlord-extensions` in the extensions load list for your overlord process. + +The extension uses the task queue to limit how many concurrent tasks (k8s jobs) are in flight so it is required you have a reasonable value for `druid.indexer.queue.maxSize`. Additionally set the variable `druid.indexer.runner.namespace` to the namespace in which you are running druid. Review Comment: So this one was a tricky one. Some users have quotas setup for their namespace while others do not. Quotas such as how many jobs can i have concurrently running. I believe if you do this, you will just have unscheduled pods while resources are not available and then when they become available they will launch. If you use aws, you could perhaps tie in the autoscaler and have some logic stating if i have x amount of pods unscheduled then scale up ec2 instances. And scale down the same with the reverse logic, if I have ec2 instances not using resources, scale things down. To answer your question, yes we should always ask users to set `maxSize` less than the k8s capacity, unless they have autoscaling configured. In that case its up to the user to determine what they can potentially scale up to. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
