Orchestrator High Availability Architecture

Hi All,

I am trying to come out with a High Availability Orchestrator setup in Prod env, like the screenshot below.


I have a few questions which would need some clarifications.

  1. To ensure HA, there will be 2 orchestrator (Prod_Orchestrator_1, Prod_Orchestrator_2) setup up in a High Availability Add-On Cluster with Load balancing . Am I right to say that both Prod_Orchestrator_1 & Prod_Orchestrator_2 will be installed on the same web server?

  2. There will also be a UiPath Studio in the Prod env. During the publish of workflow project to orchestrator, how will it work? Do I need to publish to both orchestrators to ensure that they are ‘in-sync’ with each other?

Advance thanks!

Regards,
Jack

Hi @robot_learner,

  1. You need to setup orchestrator instances on 2 different web servers.

  2. You do not need to publish to both orchestrators. As shown in the diagram, both the orchestrators can refer to same shared storage where nuget packages are stored

As you see in the diagram, robots/operator will be using NLB url to communicate with the orchestrator. When you publish it would go to the shared folder.
For robots, it would not matter how many orchestrator are used as it will be communicating using NLB url

1 Like

Hi @san.nagane,

Thanks for the reply. If the orchestrators are set up on 2 different web servers, will they be able to share the same DB? from the diagram above, it looks as if both the orchestrators are interacting with the same DB.

Thanks in advance!

Regards,
Javk

Yes… both the orchestrator connect to the same DB

1 Like

Hi @Rocki_Jan, @san.nagane,

Thanks for all the inputs. Just a quick question. This is the High Availability that was provided by UiPath

In the picture, there will be 2 orchestrators hosted on different servers. Just wanted to be certain, for such a setup, we will need to consider whether the design is a Active-Active or a Active-Passive set up too right?

Thanks in advance!

Regards,
Jack

I think it completely depends on how you want to setup. But I think for high availability it should be active active(because for HA you want all the orchestrator node up and active, so that robot can pass the data to any of the node via NLB) and for disaster recovery you can have active passive.

But again I think it depends on how do you want to setup.

Cheers

1 Like

Hi @Rocki_Jan,

Understood. Thanks!

Regards,
Jack

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.