Our developers could benefit in some instances having a Linux Desktop automatically assigned to a generic Linux server instance(agent) that could then be utilized to access other pooled development resources.
However, they also have a need for remote desktop support to specific systems. In some cases these systems are even limited to a single user utilizing them, in others there might be a group of Linux hosts where the users would want direct remote desktop to them. They may likely want to choose which host in the group that they connect to.
Thinlinc is a robust Linux Remote Desktop solution but am I missing knowledge where users have the ability using the ThinLinc client to select remote control of a specific Linux system running an agent? The documentation only appears to talk about automatically load balancing across agents. I am watching for a remote desktop solution that is robust but has to also allow manual selection of which host to attach to?
I apologize if this is a well known capability but I watched multiple videos and at least skimmed through the documentation.
It’s currently not possible for users to select an agent manually (this would defeat the purpose of having a load-balancer). But you can configure a cluster so that each user - and only that user - always ends up on the same agent.
To achieve this, you would configure one subcluster per agent, and ensure that each subcluster only allows one specific, unique user. You can read more about how to configure subclusters in ThinLinc here:
Thank you very much for the information. Thinlinc is a very compelling solution in its current form for nonspecific shared Linux Desktops.
However, I think that having Thinlinc optionally use the load balancer agent selection for different Thinklinc clusters, while also supporting the direct selection of a separate “group of non-cluster agents” would be a very compelling feature for Thinlinc.
The marketing of Thinlinc acting as a direct replacement for Omnissa VDI or NoMachine, without being able to readily have “end users” - select between a load balanced Thinlinc cluster or a specific agent on a non-cluser agent group, doesn’t really ring true. Thinlinc doesn’t really act as a replacement for direct remote desktop because in many cases for VDI or NoMachine - the end user needs to select a specific host for what that host is being used to develop.
Put another way, it would defeat the purpose to connect to an unspecific Linux Desktop instance and then simply need to use another remote desktop technology on that Linux Desktop instance to connect to a specific Linux Desktop on a development server.
Hopefully, this is food for thought and opens Thinlinc to an additionally greater market segment.
Thanks for the feedback @NyleFL, we always appreciate the end-user perspective.
What you’re describing can be achieved in ThinLinc, by setting up two clusters, each with its own master server. The user then selects whether they want a load-balanced or “dedicated” session by which cluster they connect to.
The latest version of the ThinLinc client (4.20.0, just released at time of writing) includes a new feature to help with this; it will remember previously entered server names and provide these as suggestions to the user. So managing multiple clusters in this way just got a bit easier.
Another option is to create separate client configuration files for each cluster, and use these to launch the ThinLinc client. This process is described in more detail here:
I appreciate that – that separate cluster feature is perfect, allowing separate clusters featuring different HPC, etc. technologies.
For direct Linux development where the system is specific to the developer, I wouldn’t want to have to run the server and the agent on each of the individual hosts. If that makes sense?
Just trying to paint a picture of how Thinlinc does fit but does not cover all our remote desktop requirements. For individual customized development systems that we still need remote desktop to – a non-load balanced cluster where the individual agent could be directly selected by the user would seem to meet the remainder of our need.
You wouldn’t have to. With properly configured subclusters, all users would connect to the same master machine and then be redirected to “their” agent. Each agent would only run the vsmagent service, not vsmserver.
How to do this is described in the link I posted above, but if anything is unclear then please feel free to let us know.
With many employees there is no way for us to manage all of the individual development systems to assign them manually to each developer but I did misunderstand what you were describing in your reply so thanks for clarification.
It sounds as if the missing functionality could be a future major enhancement where you could create that group of agents that allowed for manual selection of specific agent on the client.
You’re right, it’s not a solution which scales well, if you’re talking dozens/hundreds of users. As you quite correctly point out, we would need specific functionality for handling this scenario.
Just to clarify - is your requirement for one dedicated agent per user? Or did you just want the ability for users to select a specific agent, which may be shared by multiple users?
Developers would have and unknown number of individual direct Linux development systems that would require a direct connection.
Are we alone in this need? If we don’t deploy VDI for remote desktop support then just as RDP is utilized for Windows direct connections, we need a Linux solution. I understand that RHEL 10 will support RDP natively, so I could understand if Thinlinc doesn’t think investing in this area makes long term business sense. We just don’t have a good solution for RHEL 8/9 at this point and I’m not certain how stable/efficient RDP is on RHEL 10 over a cable VPN connection.
@NyleFL I think it might be worth having a quick chat. Would you mind flicking an email through to support@cendio.com, and we can arrange a time that suits?
