Code package deployment for on-premise FO environment

(0) Share

Report

Posted on by Community Member

We have two on-premise environments, sandbox and production.
We create a package of our code models from development machines and deploy it through LCS.
We deploy first to sandbox environment, release candidate and then apply that package to production.

Issue is that deploying to each environment takes several hours to full build d365 models code, creating package, preparing, downloading and updating environment via LCS considering everything goes smoothly like without any antivirus issue on file-share or any random issue that may occur during this process. So we find it difficult to do this activity more than once in a week and we schedule it at the weekend.

Is there any other way for deploying code package to on-premise environment so that we can easily deploy whenever we need? Can we automate this process? what's the standard?

Thanks in advance

I have the same question (0)

All responses (11)

Answers (3)

Verified answer

nmaenpaa 101,160 Moderator on at

Like (0)

Report

Build and creating of deployable package can be fully automated via Azure DevOps build pipeline.

Not 100% sure about on-premises, but for cloud systems you can easily automate your test deployments using ADO release pipeline. I assume it works for on-premises, too.

So the only remaining manual task is to mark the package as release candidate, and trigger the prod update.

Was this reply helpful? Yes No
Community Member on at

Like (0)

Report

Thanks Nikolaos,

Is this what what you are suggesting, check the link :
community.dynamics.com/.../release-your-dynamics-365-for-finance-and-operation-packages-to-lcs-with-azure-devops-pipelines

Was this reply helpful? Yes No
Suggested answer

nmaenpaa 101,160 Moderator on at

Like (0)

Report

Yes.

Was this reply helpful? Yes No
Verified answer

Joris dG 17,775 on at

Like (1)

Report

Sadly, LCS does not support an API for deploying to on-premises, so the Azure DevOps tasks don't work for on-prem.

Was this reply helpful? Yes No
Community Member on at

Like (0)

Report

Thanks for the reply. Is there any other way to save deployment time?

Was this reply helpful? Yes No
Suggested answer

nmaenpaa 101,160 Moderator on at

Like (0)

Report

Hi AR01,

in your original question you wrote "Issue is that deploying to each environment takes several hours to full build d365 models code, creating package, preparing, downloading and updating environment via LCS".

You listed there steps that are not related to the deployment.

But let's go through them:

- Full build D365 models, and create package: >ou only need to build your own packages, not Microsoft ones. And, the build can and should be automated. The deployable package is created automatically in the build process. So, you can set up a nightly (or afternoon) build that creates a deployable package of your latest customizations.

This way only following steps remain manual:

- Upload package to LCS and apply it to UAT - you can do this as the last task of your working day, and let the deployment do its job in the evening. This way you have the latest code in the system the next morning.

- Mark package as release candidate and apply to Prod

The On-premises package deployment performance depends on the performance of your servers, so perhaps there's something that you can improve there.

Was this reply helpful? Yes No
Community Member on at

Like (0)

Report

About "- Full build D365 models, and create package: >ou only need to build your own packages, not Microsoft ones. And, the build can and should be automated."

While we build our models from Visual studio, there is a With reference check which builds other models. You mean there is no need to mark that check?

And while creating package from Visual studio, we only see our models so mark them and package is created.

Can we automate this via ADO build pipelines? so whenever there is a check-in there is a package built automatically and we left with the manual steps.

Thank you so very much for your kind suggestions.

Was this reply helpful? Yes No
Verified answer

nmaenpaa 101,160 Moderator on at

Like (0)

Report

You don't need to build the referenced packages when building in VS. You never change them so no need to build them.

But what is more important - you should never create deployable packages using Visual Studio. Instead you should have a build VM or a hosted pipeline that builds the code that is in your source control (not the local code on your dev machine!).

Not sure if you want to build a package from every check-in, or rather have a nightly/daily build. But both are possible.

Build VM is simpler to set up than a hosted pipeline, so I recommend to start with that. Simply deploy a cloud hosted environment using topology "Build and test" from your LCS project, and it will set everything up. Then you will see the pipeline in your ADO and can configure triggers (schedules). You can also set it up on a local VM quite easily: community.dynamics.com/.../setup-a-local-build-vm

Once you are familiar with the basics, you can try to set up the hosted build pipeline, which requires no VMs: docs.microsoft.com/.../hosted-build-automation

You might also want to check my blog about build pipelines: https://community.dynamics.com/365/financeandoperations/b/365foroperationstechnical/posts/working-with-build-pipelines-build-definitions . But that's already a bit old and doesn't describe the features of the hosted build pipelines.

Was this reply helpful? Yes No
Community Member on at

Like (0)

Report

We are interested in nightly/daily build. I understand now.

Thank you so very much

Was this reply helpful? Yes No
Giovanni Laffi 15 on at

Like (0)

Report

[quote user="Joris dG"]Sadly, LCS does not support an API for deploying to on-premises, so the Azure DevOps tasks don't work for on-prem.[/quote]
Hello Joris,

may I ask you more about your answer? Is there a reference about that? I am wondering if it will work in an on-prem FinOps + on-prem ADO Server scenario.

Thank you in advance

Was this reply helpful? Yes No