I have created a simple PHP api application that works with a mysql database to store data. I have been experimenting with Kubernetes on my Windows 10 machine through Minikube.
I have just about got my head round the ideas involved, yet I’m not sure about how to implement this properly. So far I have used Kompose to create a set of yaml files from an existing docker-compose file. This has been half successful.
To get my application code into a pod hosting PHP, I have been using hostPath to share from my local machine. I mount to the minikube machine and share from there. I was having trouble sharing by other means. The application code is hosted in a github repo.
My questions are:
Is mounting my application code into a pod (assuming this is similar to what happens in docker) the correct way to do this? I’m not clear exactly what information is held on an image retrieved from the docker hub. Although I have read up on containers isolating the build environment from your machine.
How does this approach to translate into a production environment hosted on a cloud? I see there are various storage types. I had for example, wanted to try deploying on AWS just to see how this would work in practice.
I’m really looking for guidance to go from the tutorials found on the web working on my machine, to something that could be done for a customer hosted on the cloud. This might scale up to a more microservices style architecture over time.
The approach you are describing is mostly for development setups, where you want to mount your code into the container as a volume so you don't have to rebuild every time your code changes. Typically done with a docker-compose file.
For production setups, you want the docker image to correctly work and only mount volumes to data you want to persist, typically databases are the core example. For this EKS is deeply integrated into the AWS infrastructure and will create EBS volumes on demand. You don't need to provision any volume or even care for most cases (unless you need multiple read-write volumes needed for scaling).
For a PHP application you really should not persist any data in the pod, because it will create other issues when you need to scale the application. Also, a good approach for managing files that need to persist is S3 (AWS simple storage service).
So generally speaking, you need a deployment per application a service to access each pod on that application and then an ingress object to route traffic from the internet to each pod.
Your application docker image is really the core. You just build it with your code inside. Make sure to pass configuration using environment variable or configuration file so you can connect to the database.
Now for kubernetes, for each compoment (e.g. PHP application, MySQL) you will most likely create a deployment k8s manifest that points to the docker image and add some configuration environment variables.
For production, you will need persistence volume. On aws you can simply use EBS-backed volumes
To get traffic from Internet to your PHP application, you will need to add one or more k8s components:
K8s Service manifest that exposes your PHP deployment/pod on a stable address. If you only have q or very few services, you can use LoadBalancer which on cloud like AWS will create an ALB/ELB (might need to add annotation to your service)
An ingress which is just a reverse proxy (contour, nginx, traefik). On cloud environment it will map to an ALB/ELB. The advantage of this is that you can have a single ALB for all your services i.e. save money. Also you can configure routing path or TLS termination in one place.
Related
I'm trying to understand how docker works in production for a scalable symfony application.
Let's say we start with a basic LAMP stack:
Apache container
php container
mysql container
According to my research once our containers are created, we push them to the registry (docker registry).
For production, the orchestrator will take care of creating the PODS (in the case of kubernetes) and will call the images that we have uploaded to the registry.
Did I understand right ? Do we push 3 separate images on the registry?
Let me clarify a couple of things:
push them to the registry (docker registry)
You can push them to the docker registry, however, depending on your company policy this may not be allowed since dockerhub (docker's registry) is hosted somewhere in the internet and not at your company.
For that reason many enterprises deploy their own registry, something like JFrog's artifactory for example.
For production, the orchestrator will take care of creating the PODS (in the case of kubernetes)
Well, you will have to tell the orchestrator what he needs to create, in case of Kubernetes you will need to write a YAML file that describes what you want, then it will take care of the creation.
And having an orchestrator is not only suitable for production. Ideally you would want to have a staging environment as similar to your production environment as possible.
will call the images that we have uploaded to the registry
That may need to be configured. As mentioned above, dockerhub is not the only registry out there but in the end you need to make sure that it is somehow possible to connect to the registry in order to pull the images you pushed.
Do we push 3 separate images on the registry?
You could do that, however, I would not advice you to do so.
As good as containers may have become, they also have downsides.
One of their biggest downsides still are stateful applications (read up on this article to understand the difference between stateful vs stateless).
Although it is possible to have stateful applications (such as MySQL) inside a container and orchestrate it by using something like Kubernetes, it is not advised to do so.
Containers should be ephemeral, something that does not work well with databases for example.
For that reason I would not recommend having your database within a container but instead use a virtual or even a physical machine for it.
Regarding PHP and Apache:
You may have 2 separate images for these two but it honestly is not worth the effort since there are images that already have both of them combined.
The official PHP image has versions that include Apache, better use that and save yourself some maintenance effort.
Lastly I want to say that you cannot simply take everything from a virtual/physical server, put it into a container and expect it to work just as it used to.
I a want to run a web page similar like kubernetes dashboard.The web page takes input from the user and generates a small file but i want the web page to be loaded without using any server. kubernetes is deploying a pod and bringing up the web page i want to do the same.If kubernetes is also using a server how is it using it(is it directly downloading it with the OS in the pod or how is kubernetes doing it).
Overview I want to know how kubernetes dashboard is getting deployed is it using a server if so how is it getting the server installed in the kubernetes pod else how is it bring up the UI.
Actually, Kubernetes plays the role as an orchestrator and provides sufficient way for building communication channels between containers in the cluster and uses Docker by default as a container runtime.
Containers represent run-time environment for images, however images consist with OS layer and application binaries, a good explanation you can find here. In order to build own image you might consider two ways to afford this: create an image from existing one in Docker Hub or compose image from Dockerfile.To store the customized image might be the option to push it into Docker Hub repository or stand for some private isolated repo by deploying a Registry server.
When you are ready with an image, and you plan to implement application in Kubernetes cluster, that's a good time to create first microservice. Although, there are tons of materials about Kubernetes cluster and its run-time engine architecture in the globe, I would focus on the application deployment lifecycle.
Deployment is the main mechanism which defines how are Pods should to be implemented within a cluster and provides specific configuration for further application run-time workflow.
Service describes a way how the particular Pod will communicate with other resources within a cluster, providing endpoint IP address and port where your application will respond.
In general scenario with Kubernetes Dashboard, the method in use kubectl proxy will expose the application by proxying gateway between host and Kubernetes API, which is more like for testing purposes and not secure, in comparison with Nodeport type which brings more convenient way to make application accessible outside the cluster, as described in this Stack thread.
I encourage you to get some more learning stuff in the official Kubernetes documentation.
Good day
We have a development environment that consists of 6 virtual machines. Currently we are using Vagrant and Ansible with VirtualBox.
As you can imagine, hosting this environment is a maintenance nightmare particularly as versions of software/OS change. Not too mention resource load for developer machines.
We have started migrating some virtual machines to docker. But this itself poses problems around orchestration, correct configurations, communication etc. This led me to Kubernetes.
Would someone be so kind as to provide some reasoning as to whether Kubernetes would or wouldn't be the right tool for the job? That is managing and orchestrating 'development' docker containers.
Thanks
This is quite complex topic and many things have to be considered if it's worth to use k8s as local dev environment. Especially I used it when I wanted to have my local developer environment very close to production one which was running on Kubernetes. This helped to avoid many configuration bugs.
In my opinion Kubernetes(k8s) will provide you all you need for a development environment.
It gives you much flexibility and does much configuration itself. Few examples:
An easy way to deploy new version into local kubernetes stack
You prepare k8s replication controller files for each of your application module (keep in mind that they need to be stateless modules)
In replication controller you specify the docker image and that's it.
Using this approach you can push new docker images to local docker_registry and then using kubectl control the lifecycle of your application.
Easy way to scale your application modules
For example:
kubectl scale rc your_application_service --replicas=3
This way k8s will check how many pods you have running for your service and if it recognises that the number is smaller then the replicas value it will create new to satisfy the replicas number.
It's endless topic and many other things come to my mind, but I would suggest you to try it out.
There is a https://github.com/kubernetes/kubernetes/blob/master/docs/devel/developer-guides/vagrant.md project for running the k8s cluster in vagrant.
Of course you have to remember that if you have many services all of them have to be pushed to local repository and run by k8s. This will require some time but if you automate local deploy with some custom scripts you won't regret.
As wsl mentioned before, it is a quite complex topic. But i'm doing this as well at the moment. So let me summaries some things for you:
With Kubernetes (k8s) you're going to orchestrate your SaaS Application. In best case, it is a Cloud-native Application. The properties/requirements for a Cloud-native Application are formulated by the Cloud Native Computing Foundation (CNCF), which basically were formed around k8s, after Google donates it to the Linux Foundation.
So the properties/requirements for a Cloud-native Application are: Container packaged, Dynamically managed and Micro-services oriented (cncf.io/about/charter). You will benefit mostly from k8s, if your applications are micro-service based and every service has a separate container.
With micro-service based applications, every service can be developed independently. The developer only needs to follow the 12Factor Method (12factor.net) for example (use env var instead of hard coded IP addresses, etc).
In the next step the developer build the container for a service and pushes it the a container registry. For a local develop environment, you may need to run a container registry inside the cluster as well, so the developer can push and test his code locally.
Then you're able to define your k8s replication-controllers, services, PetSets, etc. with Ports, Port-mapping, env vars, Container Images... and create and run it inside the cluster.
The k8s-documentation recommend Minikube for running k8s locally (kubernetes.io/docs/getting-started-guides/minikube/). With Minikube you got features like DNS, NodePorts, ConfigMaps and Secrets
Dashboards.
But I choose the multi node CoreOS Kubernetes with Vagrant Cluster for my Development Environment as Puja Abbassi mentioned in the Blog "Finding The Right Local Kubernetes Development Environment" (https://deis.com/blog/2016/local-kubernetes-development-environment/), it is closer to the my production environment (12Factor: 10 - Dev/prod parity).
With the Vagrant Environment you got features like:
Networking with flannel
Service Discovery with etcd
DNS names for a set of containers with SkyDNS
internal load balancing
If you want to know, how everything works look inside this Github repo github.com/coreos/coreos-kubernetes/tree/master/multi-node (vagrant and generic folder).
So you have to ask yourself, if you or your developers really need to run a complete "cloud environment" locally. In many cases a developer can develop a service (based on micro-services and containers) independently.
But sometimes it is necessary to have multiple or all services run on your local machine as a dev-environment.
I couldn't have enough courage to start using docker now I'm feel like came from last century. I want to clear my doubts about docker before get started. My question is mainly for deploying/running docker images on cloud or hosting environment.
Can I build a docker image with any type of server (eg. wildfly, payara) and/or database server (eg. mysql, oracle) and will it work on docker enabled cloud/datacenter?
If it's yes how about persistent datas like database files and static storages (eg. images, uploaded documents, logs) those are stored in docker images or somewhere else? What will happen to those files when I update my application and redeploy new image?
I read posts about what is docker but I couln't find specific answer. Forgive me for not doing enough googling.
I have run docker on AWS and other cloud providers. It is really not that hard if you have some experience with system administration and or devops. Regarding cloud hosters and getting started, most providers have some sort of tutorial on how to get started using docker with their infrastructure:
http://docs.aws.amazon.com/AmazonECS/latest/developerguide/docker-basics.html
https://azure.microsoft.com/en-us/documentation/articles/virtual-machines-linux-dockerextension/
Can I build a docker image with any type of server (eg. wildfly,
payara) and/or database server (eg. mysql, oracle) and will it work on
docker enabled cloud/datacenter?
To get a server up and running, you just need the docker engine installed on the host, there are packages for many distros:
https://docs.docker.com/engine/installation/
After docker engine is installed, you can create dockerfiles for basically any server or service. Hopefully you do not need to, in most cases, since there are countless docker files and pre-configured, vendor maintained images already available on dockerhub (I use wildfly, elk-stack, and mysql for example). Be careful about selecting images are maintained, otherwise you end up with security issues in your images that might never get fixed! Or you have to do it yourself!
Example images:
https://hub.docker.com/r/jboss/wildfly/
https://hub.docker.com/_/mysql/
https://hub.docker.com/_/oraclelinux/
https://hub.docker.com/u/payara/
If it's yes how about persistent datas like database files and static
storages (eg. images, uploaded documents, logs) those are stored in
docker images or somewhere else? What will happen to those files when
I update my application and redeploy new image?
In general, you will want to store persistent data external to the docker image and mount it into the image as a volume:
https://docs.docker.com/engine/tutorials/dockervolumes/
Some cloud based storage providers might be easier to mount or connect to in other ways, but this volume approach is standard, IMO.
For logfiles, I actually push them to an ELK server, so having a volume for the logs is not necessarily required. However, since the ELK server is also a docker image, it does have a volume where the data is persisted.
So you have:
documentation from your cloud hoster (or docker themselves)
a host in your cloud running docker engine
0..n images that you can either grab from dockerhub or build yourself.
storage for persistent data on this host or mounted from elsewhere that you mount into your docker images on startup. this is where e.g. mysql data folders live, or where you can persist logs, etc.
Of course, it can get much more complex from there, e.g. how to transparently scale and update your environment etc., but that is something for e.g. kubernetes or docker swarm or some other solution (I've scripted a bit on my own but do not need the robustness or elastic scalability of large systems).
Regarding cluster management, it should be noted that Swarm is now included in the Docker Core. This has created some controversy in the community and even talks of a fork of the core:
https://technologyconversations.com/2015/11/04/docker-clustering-tools-compared-kubernetes-vs-docker-swarm/
https://jaxenter.com/docker-1-12-is-probably-the-most-important-release-since-1-0-129080.html
http://searchitoperations.techtarget.com/news/450303918/Docker-fork-talk-prompts-container-standardization-brawl
http://www.infoworld.com/article/3118345/cloud-computing/why-kubernetes-is-winning-the-container-war.html
I have experience running docker on Alibaba cloud and AWS as well. I did not see any difference in working with docker on both cloud providers. Docker images can be build same way on all linux platform regardless of the cloud provider. However, persistence of data need to be taken care using docker volumes. However, it is recommended to use managed service such as RDS in Alibaba cloud for databases instead of using docker.
Can I build a docker image with any type of server (eg. wildfly,
payara) and/or database server (eg. mysql, oracle) and will it work on
docker enabled cloud/datacenter?
You can build your own Docker images or use solutions that are already pre-packaged and proven by cloud providers. For example, here is an auto-clustering Docker-based implementation of GlassFish that can be run and managed on Jelastic PaaS.
If it's yes how about persistent datas like database files and static
storages (eg. images, uploaded documents, logs) those are stored in
docker images or somewhere else? What will happen to those files when
I update my application and redeploy new image?
With the above mentioned cluster, all data is kept inside containers and stays without changes after restart. As an option, you can also connect a separate data storage container if you wish to share it across other containers.
I’m using Amazon ECS to auto deploy my containers on uat/production.
What is the best way to do that?
I have a REST api with a several front-end clients
Should I package my api container with nginx in the same container?
And do the same thing with the others front end clients.
Or I have to write a big task definition to bring together all my containers(db, nginx, php, api, clients) :(, but that's mean that I should redeploy all my infrastructure at each push uat/prod
I'm very confusing.
I would avoid including too much in a single container. Try and distill your containers down to one process doing one thing. If all you're doing is serving up a REST API for consumption by your front end, just put the essential pieces in for that and no more.
In my experience you also want your ECS tasks to be able to handle failure gracefully and restart, and the more complicated your containers are the harder this is to get right.
Depending on your requirements I would look into using ELB instead of nginx, you can have your ECS cluster point at an ELB and not have to deal with that piece at all.
Do not use ECS - it's too crude. I was using it as a platform for our staging/production environments and had odd problems during deployments - sometimes it worked well, sometimes - not (with the same Docker images). ECS provides not clear model of container deployment and maintenance.
There is another good, stable and predictive option - Docker Cloud service. It's new tool (a.k.a. Tutum) that was acquired by Docker. I switched the CI/CD to use it and we're happy with it.
Bind Amazon user credentials to Docker Cloud account. Docker Cloud uses AWS (or other provider) API for creating appropriate computer instances.
Create Node. Select Amazon EC2 instance type and parameters of storage, security group and so on. New instance will contain installed docker software and managing container that handles messages from Docker Cloud (deploy, destroy and others).
Create Stackfile, see https://docs.docker.com/docker-cloud/apps/stack-yaml-reference/. Stackfile is a definition of container group you required. You can define different scaling/distribution models for your containers using specific Stackfile options like deployment strategy, see https://docs.docker.com/docker-cloud/apps/stack-yaml-reference/#deployment-strategy-1.
Define ELB configurations in AWS for your new instances.
P.S. I'm not a member of Docker team and I like other AWS services :).
Here is my two cents on the topic, the question is not really related to ecs, it applies to any body deploying their apps on docker.
I would suggest separating the containers, one for nginx and one for API.
if they need to be co-located on the same instance, on ECS you can define them as part of the same task and on kubernetes you can make them part of same pod.
Define a docker link between the nginx and the api container. This will allow the nginx process to talk to api container without the api container exposing its ports to the host.
One advantage of using the container running platforms such as kubernetes and ecs is that they ensure each of the container run all the time and dynamically restart if one of the processes/containers go down.
Separating the containers will allow these platforms to monitor both the processes separately. When you combine the two into one container the docker container can only run with one of the processes in foreground, so you will loose the advantage of auto-healing for one of the processes.
Also moving from nginx to ELB is not a straightforward solution, you may have redirections and other things configured on the nginx, which are not available on ELB(As of date).
If you also need the ELB, there is no harm in forwarding the requests from the ELB to the nginx port.