Per-node resource allocation view in PDM / Proxmox cluster

[Pabz9]

Hello,


I have a question that has been bothering me for some time.


I am currently running a 6-node Proxmox cluster with a total of 4 TiB of RAM and 672 CPU cores. Since my nodes are very heterogeneous in terms of hardware capacity, I often need to rebalance my most demanding guests across the cluster.


Some nodes have a lot of RAM but fewer CPU cores, some have the opposite profile, and others are more balanced. Because of this, I need to know, for each node, how many resources are currently allocated and how that compares to the node’s total capacity.


Until now, I used a script that queried the cluster with pvecm and then populated an Excel sheet with the results.


I recently discovered PDM and thought it might provide this kind of visibility out of the box. From what I can see, the information does exist, but it is only displayed as an aggregate view for the whole cluster, not node by node.


So I am wondering:

• Is there a way in PDM to display allocated resources per node rather than only at the cluster level?
• If not, is there another recommended way in Proxmox to retrieve this information cleanly for each node?
• Also, is there a specific reason why this information is only presented in an aggregated cluster-wide view, even though the underlying per-node data seems to exist?

Thank you in advance for your help.

 

[UdoB]

I cannot answer your question, sorry. I just want to mention this project: https://github.com/credativ/ProxLB

From my limited(!) point of view PDM is meant to be helpful when managing multiple and independent clusters (or multiple single hosts).

 

[Pabz9]

Thank you for your reply.


I know that PDM is not really meant for managing or monitoring only a single cluster. Still, I find it useful because, in addition to my main cluster, I also have a PBS instance and another standalone Proxmox node, so having everything centralized is very convenient.


The reason I asked is that i knew that in Proxmox VE, the allocated resources vs total resources metric is not available in the GUI. I had already seen a few threads mentioning this when I looked into it before.


What surprised me is that this metric now appears to exist in PDM, although only as a cluster-wide aggregate. That is why I was wondering why it cannot also be shown per node, since the underlying data seems to be there already.


I understand that this may not be a particularly important metric for most users, which is probably why Proxmox has not focused on exposing it.


Thank you also for sharing the load balancer project. I did not know about it, and I will definitely look into it. I am not sure whether I can use something like that anytime soon, since my infrastructure is not yet ideal for hyperconvergence for several reasons: different CPU types across nodes, some VM disks on Ceph and others not, and some hardware passed through to VMs via PCI.


Still, I think it could become very relevant as my infrastructure evolves.