Exploring Proxmox Datacenter Manager Alpha: First Impressions
ฝัง
- เผยแพร่เมื่อ 10 ก.พ. 2025
- Proxmox has recently recently released Proxmox Datacenter Manager in alpha. This new tool lets you manage and view the status of multiple PVE nodes in one web page. In this video I take a look at the tool as it is now in alpha and talk about the roadmap. I also share my thoughts on its state now and their roadmap for the future.
As a vmware admin of clusters globally and as we transition to Proxmox, having one "vcenter" type management interface to manage all proxmox clusters globally is essential and critical for easier visibility and management.
you can only "manage" it like stop vms backup migrate and other than that its a glorified bookmark page with statistic. On their roadmap they have no plans to expand the functionality beyond these. One can only hope but if you see the interface its looks like more of the proxmox backup interface than the original fully functioning node/cluster interface. Its a shame that their directive is different than a management interface.
Proxmox doesn't play nice if any nodes go down at all on the cluster. You have to deal with that immediately. That would be quite problematic if you were in the middle of something critical then had to stop that because your backup redis node went down lol
Proxmox Dev here, great Video ;)
Regarding the Information density: You can actually change the theme & spacing in the 'user menu' on the top right. Just wanted to put that info out there
Glad to hear a dev are watching my videos.
Thanks for pointing out those options, I'm glad the different view sizes have been added, and I like the options that are available for UI customization.
Thanks!
Thanks for the support. Glad you like the video.
Thanks B, you make some very valid points, Many of us wont need this for home use.
Getting better and better...
This is going to be a sweet tool to have for large deployments and multiple clusters. I have noticed when adding another remote I can't use the same token name so have to use a different one which is fine as each node will have it's own API token. Hopefully they will make it easier to add remotes without going through multiple steps.
Biggest feature I hope to see is fully customizable dashboard. Even make use of widgets. Can't wait to see what it looks like in six months.
Yea that seems like it would work from the error message I saw. I just wish it had an auto fail over to a different node from the cluster if possible.
When you add 2 nodes from one cluster that way is count of VMs and nodes on the dashboard correct or are some items double counted?
Hello dear EW. I think you should send your ideas regarding improving this directly to ProxMox!
If they wanna compete in the data center, I feel like they need to get the networking story right. It's still fiddly the last time I tried a year or 2 ago and requires basically manually provisioning bridges / ovs. The thing that made vmware so nice (with the vsx / Nicira stuff) is that it just set up the SDN layer so easily. Definitely want to see how this evolves.
There working a good amount on their SDN solution and that should help a lot with those issues. I should do a video on the SDN features sometime. Its not super well documented and there still seems to be a good amount work being done by Proxmox with networking.
They REALLY need to fix the NIC names changing out from under you when a kernel updates. I pin my kernel version when it works
Thats literally what i thought it would do if i cluster 2 nodes xD
One annoyance with Proxmox is it doesn't like 2 node clusters as everything stops if a node dies. The one upside with something like PDM is you can still use one node if the other was to go offline.
I hope their idea of SDN is more than "have a network between hosts". That fills the most common need, but only shares the N with SDN, and would be a setback for long term. We need at least support for real SDN use cases in the vm stack to not close doors for the future. Like security rules on the network, defining topology etc. The 2013ish real idea of SDN. Why? Most of all ransomware, national level IT defense progress, reliability, efficiency.
From the release notes I saw there was no details about how SDN would be implemented between nodes, so its a wait and see it seems. I agree with you that I hope they implement SDN well here.
I don't really understand how it works with clusters - can you only add 1 node from a cluster as a 'remote' that you can view in PDM? Like what you said about how you couldn't add the machine because it was already part of a cluster that had been added
It looks like you can only add one node in a cluster(there may be a way to add an additional time to bypass this limit though). Then once the cluster is added you can view the status of all the VMs and nodes in the cluster.
@@ElectronicsWizardry I think you can manually add more nodes to a remote entry, at least it looks like that if you edit the remote entry in PDM or add a new remote in PDM (Tab "endpoints")
Is there any impact when migrating VMs between non clustered nodes?
I did t notice any impact. The VMs worked as expected on the different nodes. Extra configuration may be needed for advanced vm configs like pcie passthrough or other host specific features but I’d generally expected migrated VMs to just work.
does it work as a central API endpoint for all my nodes - I run 2 x standalone nodes multiple places
There seems like there will be a API for PDM that should allow you to manage all the nodes. But I don't think the full docs are out yet with what the API can do.