In a microservices architecture what is the best strategy for keeping many developer environments up-to-date across multiple source code repositories?
Suppose there were 10 teams of 10 developers working on 200 microservices in git. Every developer would need to pull regularly from every repository. This could be done with scripts but is there a better way? Are we doing this wrong because it seems like a heavy overhead.
I wouldn't advise having every developer build every microservice. I would propose some sort of continuous integration environment. One centralized build server connected to all of the git repos.
Each time a repo is updated the build server should detect the change, build the code, run unit (and or functional) tests, and then push the service to some sort of integration environment. The build server may then also run some integration testing against the deployed service.
Most developers should be able to do all their development and test without needing access to the other microservices. If a developer is building service X, which depends on Y & Z and is depended on by A & B then the developer should, for the most part, only have service X. For unit testing services Y & Z should be mocked/simulated.
The challenge is going to be preventing the developer from breaking services A & B by making a change to service X. That sort of integration testing tends to be trickier as developers working on service X often don't know details (or even how to use) upstream services (e.g. A & B).
The way to tackle that I believe would be to have regular integration testing, either triggered by the build of service X or run on a regular basis. With a project this complicated a strong and robust unit test philosophy and integration test framework is going to be essential.
This comes down to good your service location/detection technology is. All one service needs to know is where to send a request X in order to get a response Y from a particular service. If this is well implemented. you could very well have each developer run whatever subset of services he is directly working on locally and have the rest of the services run in a common environment (lets call it Remote)
Remote will be configured to run all services in your platform and will have some way to get the latest code (based on cadence or based on regular intervals). Local environments could be configured to that all the services running locally will know what else is running locally and will know how to reach to the Remote services when they need any information. For developer environments, add conventions where each developer run minimum number of services required for productive development. This means if you are not working the code a service, run it remotely so you know you are running the latest code and are not out of date.
Couple of gotchas with this approach
you could run service A locally and have service B running on remote. you test everything out locally and it seems to be working so you decide to push. While you are pushing, another developer also pushes changes to B and now your changes are not longer compatible.
If you are running services A and C locally and B remotely, it should be pretty straight forward to route requests to service B to the remote environment. You should however be careful that if A calls B and B calls C then the call to C needs to be routed from the remote environment to your local C service and not to your Remote C service.
Testing - You can get over a lot of issues related to testing with a complex environment like this by have two separate suite of tests. 1. unit tests- tests that test individual components in your service with all calls to other service being mocked. 2. environment integration tests- these tests validate the communication between different services. Suite 1 will check the internal code of your service and Suite 2 will run after you push your changes to remote and will continuously ensure that the inter-service communication is as expected.
Hope that helps
Related
I work with teams members to develop a microservices architecture but I have a problem with the way to work. Indeed, I have too many microservices and when I run them during my development, it consumes too memory even with a good workstation. So I use docker compose to build and execute my MSA but it takes a long time. One often hears about how technically build an MSA but never about the way to work efficiently to build it. How do you do in this case ? How do you work ? Do you use tools or any others to improve and facilitate your developments. I've heard about skaffold but I don't see what the difference is with docker compose or with a simple ci/cd in a cluster env for example. Feel free to give tips and your opinion. Thanks
I've had a fair amount of experience with microservices and local development and here's been some approaches I've seen:
Run all the things locally on docker or k8. If using k8, then a tool like skaffolding can make it easier to run and debug a service locally in the IDE but put it into your local k8 so that it can communicate with other k8 services. It works OK but running more than 4 or 5 full services locally in k8 or docker requires dedicating a substantial amount of CPU and memory.
Build mock versions of all your services. Use those locally and for integration tests. The mock services are intentionally much simpler and therefore easier to run lots of them locally. Obvious downside is that you have to build mock version of every service, and you can easily miss bugs that are caused by mock services not behaving like the real service. Record/replay tools like Hoveryfly can help in building mock services.
Give every developer their own Cloud environment. Run most services in the cloud but use a tool like Telepresence to swap locally running services in and out of the cloud cluster. This eliminates the problem of running too many services on a single machine but can be spendy to maintain separate cloud sandboxes for each developer. You also need a DevOps resource to help developers when their cloud sandbox gets out of whack.
Eliminate unnecessary microservice complexity and consolidate all your services into 1 or 2 monoliths. Enjoy being able to run everything locally as a single service. Accept the fact that a microservice architecture is overkill for most companies. Too many people choose a microservice architecture upfront before their needs demand it. Or they do it out of fear that they will need it in the future. Inevitably this leads to guessing how they should decompose the system into many microservices, and getting the boundaries and contracts wrong, which makes it just as hard or harder to fix in the future compared to a monolith. And they incur the costs of microservices years before they need to. Microservices make everything more costly and painful, from local development to deployment. For companies like Netflix and Amazon, it's necessary. For most of us, it's not.
I prefer option 4 if at all possible. Otherwise option 2 or 3 in that order. Option 1 should be avoided in my opinion but it is probably the option everyone tries first.
In GKE and assuming you have a private cluster. You can utilize port forwarding while hooked up to the GKE environment through the CLI. Create a script that forwards your local ports to the GKE environment. I believe on the services tab in your cluster is where you will find the "port-forwarding" button that will give you the CMD command. This way you can work on one microservice with all of its traffic being routed to the actual DEV cluster. This prevents you from having to run multiple projects at the same time.
I would say create a staging environment which will have all services running. This staging environment will specifically be curated for development. E.g. if it's deployed using k8s then you expose some ports using nodeport service if you need them for your specific microservice. And have a DevOps pipeline to always keep this environment up to date with the code.
This environment should always be built from master branch. If you have single repo for app or repo per service, it's fair assumption that the will always have most recent code when you create your dev/feature branch.
Then when you want to develop a feature or fix a bug you checkout your microservice. And if you are following the microservice pattern appropriately, that single microservice should be an executable and have it's own docker file and should be debuggable from your local IDE. Many enterprises follow this pattern, and enforce at the organization level that the master branch is always production ready and high quality.
Let's say, you discover a bug in some other microservice running in k8s cluster. You will very likely get tempted to find a way to debug that remote microservice. However, that should be written as a bug for the team that owns the microservice. If your team owns it then you fix it and then start working on your feature. If you really think you need to debug multiple microservices, then I think you have real tight coupling between the services or you don't really need the microservice architecture.
I'm working on a website using PhpStorm. For a long time I developed it locally, but then I got hosting and a remote ftp server.
I created a new project in PhpStorm with the settings for remote host, and I found that deploying code takes long time (over a minute) before I can see the result, which is quite uncomfortable when debugging.
Is there any possibility to work with code on a local server, and, when I think that the project is ready for deploy, just send it to the server.
I understand, that I can just work in two different projects and just deploy the "ready" version to server via FTP, but maybe there is some more comfortable way?
There is several answers to this question, and most of them opinion based but i will try and keep it objective.
Case 1
A big corporation gives every developer a sandbox, to test their code from, the corp requires every developer to keep their code on the sandbox.
Using mounted drives could be extremely slow. Especially when PhpStorm is indexing.
Case 2
An easy way to keep an auto backup of your code it to use the build in (s)ftp(s) upload/deploy.
Solution
In both cases you could use the auto deploy feature that saves every changes to the server, that way the deploy doesn't take over a minute, but is usually already there before you know it.
I cannot recommend to use the deployment for Production as it will not pass through your version control, SAT, security setups etc. In that case I would suggest something like rocketeer etc.
EDIT:
As for 2 projects, well you can define 2 different deployment servers, and use the default one for your testing, with auto upload or something, and then the other one can be selected from the deployment menu.
What is the best way to achieve DevOps with XPages.
Multiple Developers working as a team, On Premises Servers [Dev, QA,
Prod] can we replicate to Bluemix? Source Control Automated Testing UI
/ Application, Unit testing business logic with testing framework, Automated Deployment
IDE/Tools
Domino Designer; are there other ways?
Note: Use of Views when the data is in a NSF, otherwise data is in the cloud, or SQL. No Forms or other classic Notes design elements.
What are your approaches to this?
This is a high level overview of the topics required to attempt what you're describing. I'm breezing past lots of details, so please search them out; I've tried to reference what I'm currently aware of as far as supporting documentation and blog posts, etc. of others. If anyone has anything good to add, I'm happy to add it in.
There are several components involved with what you're describing, generally amounting to:
scm workflow
building the app (NSF)
deploying the built app to a Domino server
Everything else, such as release workflow through a QA/QC environment, is secondary to the primary steps above. I'll outline what I'm currently doing with, attempting to highlight where I'm working on improving the process.
1. SCM Workflow
This can be incredibly opinionated and will depend a lot on how your team does/wants to use source control with your deployment / release process. Below I'll touch on performing tests, conceptually, during/around the build step.
I've switched from a fairly generic scm server implementation to a GitLab instance. Even running a CE instance is pretty fantastic with their CI runner capabilities. Previously, I had a Jenkins CI instance performing about the same tasks, but had to bake more "workflow" into the Jenkins task, whereas now most of that logic is in a unified script, referenced from a config file (.gitlab-ci.yml). This is similar to how a Travis CI or other similar CI config file works.
This config calls some additional helper work, but ultimately revolves around an adapted version of Egor Margineanu's PowerShell script which invokes the headless DDE build task.
2. Building an NSF from Source
I've blogged about my general build process, with my previous Jenkins CI implementation. I followed the blogging of Cameron Gregor and Martin Pradny for this. Ultimately, you need to:
configure a Windows environment with Domino Designer
set up Domino Designer to import from ODP (disable export), ensuring Build Automatically is enabled
the notes.ini will need a flag of DESIGNER_AUTO_ENABLED=true
the Jenkins CI or GitLab CI runner (or other) will need to run as the logged in user, not a Windows service; this allows it to invoke the "headless dde" command correctly, since it runs in the background as opposed to a true headless invocation
ensure that Domino Designer can start without prompting for a user's password
My blog post covers additional topics such as flagging the build as success or failure, by scanning the output logs for being marked as a failed build. It also touches on how to submit the code to a SonarQube instance.
Ref: IBM Notes/Domino App Dev Wiki page on headless designer
Testing
Any additional testing or other workflow considerations (e.g.- QA/QC approval) should go around the build phase, depending on how you set up your SCM workflow. A lot of the implementation will revolve around the specifics of your setup. A general idea is to allow/prevent deployment based on the outcome of the build + test phase.
Bluemix Concerns
IBM Bluemix, the only PaaS that runs IBM XPages applications, will require some additional consideration, such as:
their Git deploy process will only accept a pre-built NSF
the NSF must be signed by the account owner's Bluemix ID
Ref:
- IBM XPages on Bluemix
- Bluemix Docs: Building XPages apps for the Bluemix Runtime
3. Deploy
To Bluemix
If you're looking to deploy an XPages app to run on Bluemix, you would want to either ensure your headless build runs with the Bluemix ID, or is at least signed with it, and then deploy it for a production push either via a git connection or the cf/bluemix command line utility. Bluemix's receive hooks handle all the rest of the deployment concerns, such as starting/stopping the server instance, etc.
To On-Premise Server
A user ID with appropriate level credentials needs to perform the work of either performing a design replace/refresh or stopping a dev/test/staging server, performing the file copy of the .nsf, then starting it back up. I've heard rumors of Cameron Gregor making use of a plugin to Domino Designer to perform the operations needed for OSGi plugin development, which sounds pretty useful. As most of my Domino application development is almost purely NSF based, I'm focusing more on an approach of deploying to a staging/dev/test server, which I can then perform a design task on to do the needed refresh/replace; closer to the "normal" Domino way of doing things.
Summary
Again, there are a lot of moving pieces involved here, some of which gets rather opinionated rather quickly. For example, I'm currently virtualizing my build machine, so I can spool up a couple virtual machines of it, allowing for more than one build at a time. If there are major gaps in the process, let me know and I'll fill it what I can.
I have introduced branching/merging to my team and have talked before about how it would be great to automatically build and deploy code checked into the staging/master branches, but I'm a junior dev, not very ops-y.
The trouble I'm having, is that we create intranet applications and store them on our own VM's which we have access to, but we also have load balancing which is causing me grief!
I can get a build to automate (well, I haven't got all the bugs figured out but I'm working my way through them) - and I can even get the build to automatically create a zip file ready for deployment.
Is it possible to configure several servers for deployment?
I.E
1) I check in some code to stage
***Automatically***
2) Code builds
3) Build completes, Unit tests run and they complete
4) Code is packaged into a .zip
5) .Zip is deployed across the three load balancing servers (all with the same file path).
***
Maybe worth noting we currently have our TFS server running Visual Studio so the code is built on the same server it is all stored, but this is not the server we run live code from.
Any help or tutorials specific to my setup would be GREATLY appreciated, I really want to turn this departments releasing strategies around!
I am going to address only the deployment aspect. There are a lot of different ways that this can be handled, such as:
Customizing the build template
Writing custom .Net code and inserting it into the build template (which would also involve customizing the template)
Creating a Batch or Powershell script set to run after the build completes
Using a separate tool such as OctoDeploy or Release Manager to handle the deployments
The first thing you need to do is separate the build and deployment steps in your head. While they are tightly coupled in your model, they are two totally different tasks that need to be handled different ways.
The second thing is to stop thinking like a developer when it comes to the deployment portion. While there will likely be a programmatic solution, you'll need to identify the manual steps first.
You stated that you're not very ops-y, by which I assume you mean you're more Developer and not Systems Analyst. If that is the case, then the third thing you'll need to do is get someone who is involved, such as your current release team.
There are 3 major things that need to be done then:
EVERYTHING needs to be standardized. If you can't standardize something, then standardize the way that it's non-standard (example: You have a bulk list of servers you need to deploy to, and you need to figure out which ones to deploy to based on their name, which can be anything. In that case, a rule needs to be put in place that all QA servers need to have QA in their name, User Acceptance servers need UAT, Production need PROD, etc.).
Figure out how you're going to communicate from the build to the deployment, which builds are going to deployed, to which servers, and where the code is going to be picked up from
You need to document every manual step, and every exception to those steps, and every exception to those exceptions.
Once you have all those pieces in place, you need to then go through each manual step and automate it, whether that's through Batch, Powershell, or a custom-built application. Once you have all the steps automated, you'll have both the build and deploy pieces complete.
After you're able to execute a single "manual" automatic deployment to a single environment, you're then ready to figure out how you want to run it for multiple environments. This can be as complex as an XML file that is iterated through, to simply calling the same command multiple times with different parameters.
A quick summary of how I've done this at my current job (where using a third-party deployment tool was not an option):
Created a tool using .Net WinForms to allow us to "manually" run automated builds (We use the interface to determine the input parameters, and the custom classes under the hood do all the heavy lifting. These custom classes are in a separate project that builds to their own dll. This also allows us to test tweaks and changes to the process in a testing environment before we roll it out to our production build server)
Set up an XML file for each set of environment (QA, UAT, Prod, etc.) that contains all of the servers that need to be deployed to in that environment, including destination paths, scheduled tasks, and Windows Services
Customize the TFS build template and include the custom classes created for the custom tool, which will read the XML file and iterate through each server entry to perform the deployments
I'm more than happy to help with more specific examples and assistance, I look at things a bit different than most people and it helps when it comes to release management.
I am setting up an automated deployment environment for a number of decoupled services that are in active development. While I am comfortable with the automated deployment/configuration management aspect, I am looking for strategies on how best to structure the deployment environment to make things a bit easier for developers. Some things to take into consideration:
Developers are generally building web applications, web services, and daemons -- all of which talk to one another over HTTP, sockets, etc.
The developers may not have all running on their locally machine, but still need to be able to quickly do end to end testing by pointing their machine at the environment
My biggest concern with continuous deployment is that we have a large team and I do not want to constantly be restarting services while developers working locally against those remote servers. On the flip side, delaying deployments to this development environment makes integration testing much more difficult.
Can you recommend a strategy that you have used in this situation in the past that was worked well?
Continuous integration doesn't have to mean continuous deployment. You can compile/unit test/etc the code "continuously" thoughout the day without deploying it and only deploy at night. This is often a good idea anyway - to deploy at night or on demand - since people may be integration testing during the day and wouldn't want the codebase to change out from under them.
Consider, how much of the software can developers test locally? If a lot, they shouldn't need the environment constantly. If not a lot, it would be good to set up mocks/stubs so much more can be tested on a local server. Then the deployed environment is only needed for true integration testing and doesn't need to be update constantly throughout the day.
I'd suggest setting up a CI server (Hudson?) and use this to control all deployments to both your QA and production servers. This forces you to automate all aspects of deployment and ensures that the are no ad-hoc restarts of the system by developers.
I'd further suggest that you consider publishing your build output to a repository manager like Nexus , Artifactory or Archiva. In that way deployment scripts could retrieve any version of a previous build. The use of a repository manager would enable your QA team to certify a release prior to it's deployment onto production.
Finally, consider one of the emerging deployment automation tools. Tools like chef, puppet, ControlTier can be used to further version control the configuration of your infrastructure.
I agree with Mark's suggestion in using Hudson for build automation. We have seem successful continuous deployment projects that use Nolio ASAP (http://www.noliosoft.com) to automatically deploy the application once the build is ready. As stated, chef, puppet and the like are good for middle-ware installations and configurations, but when you need to continuously release the new application versions, a platform such as Nolio ASAP, that is application centric, is better suited.
You should have the best IT operation folks create and approve the application release processes, and then provide an interface for the developers to run these processes on approved environments.