Proxmox Datacenter Manager - First Alpha Release
- Thread starter t.lamprecht
- Start date
Good start and now i have 1 place where i can monitor all my pve hosts. caught a bug and posted it separately.
I installed PDM today and tested it across several clusters. My primary clusters consist of at least 10 hosts, but for the Alpha test, I added three clusters. Here are my findings so far:
1. I frequently encounter the error: "NetworkError when attempting to fetch resource".
2. When moving between clusters, I receive the message: "VM xxx already exists". It seems that fetching the next free VM ID does not work correctly in the target cluster during migrations.
3. There appears to be an issue with remote web links. All my clusters use FQDNs, but the links to the web interfaces or shells only display hostname:8006.
Suggestion for Improvement:
It would be great if folders or locations could be created under the “remotes” section. We use multiple clusters per physical location, and in some cases, standalone PVEs are used at these locations due to licensing requirements. A folder structure with 2 or 3 levels would work perfectly for organizing this setup.
Let me know if further details or logs are required to assist in addressing these issues.
1. I frequently encounter the error: "NetworkError when attempting to fetch resource".
2. When moving between clusters, I receive the message: "VM xxx already exists". It seems that fetching the next free VM ID does not work correctly in the target cluster during migrations.
3. There appears to be an issue with remote web links. All my clusters use FQDNs, but the links to the web interfaces or shells only display hostname:8006.
Suggestion for Improvement:
It would be great if folders or locations could be created under the “remotes” section. We use multiple clusters per physical location, and in some cases, standalone PVEs are used at these locations due to licensing requirements. A folder structure with 2 or 3 levels would work perfectly for organizing this setup.
Let me know if further details or logs are required to assist in addressing these issues.
Last edited:
the one most annoying thing that's being a constant in versions of proxmox since like version 5 is migration tries to copy the disk on the root first and then restores it. please stop doing that and instead directly create the vm/lxc disk.
Besides that, great job! we needed something that made managing multiple nodes non-cluistered easier and i appreciate the hard work that went into this.
Besides that, great job! we needed something that made managing multiple nodes non-cluistered easier and i appreciate the hard work that went into this.
Good note. I imagine starting a tmux in the web shell will protect you from this problem.
Cluster is for HA. This solution doesn't change the HA concept. The cluster to cluster migration is allready in PVE for a long time "qm remote-migrate". If you use clustering now for single-point-of-management than this can replace that.
I get the same 501 issue and I used the password method. Did you install PDM from ISO or did you install via the repository? Because I installed on a Debian Bookworm system using the repository.
Found a bug when using Safari browser on MacOS.
When joining a new cluster - after verifying communication (which failed to pull/verify my public certificate, which is strange, but I assume it is because it is connecting with a shortened domain name, despite that I types in the FQDN... But I just added the fingerprint and went on) - when I on tab 2 "Settings" should type in realm, it doesn't show any realms..... In a Chrome browser I can easily select "PAM"....
When joining a new cluster - after verifying communication (which failed to pull/verify my public certificate, which is strange, but I assume it is because it is connecting with a shortened domain name, despite that I types in the FQDN... But I just added the fingerprint and went on) - when I on tab 2 "Settings" should type in realm, it doesn't show any realms..... In a Chrome browser I can easily select "PAM"....
Currently, we only connect to notes that are configured for a remote. This allows admins to explicitly select a smaller set of nodes to, e.g., create tunneling for or expose on a wider network. So if you have a 5 node cluster but only told PDM to use one of that nodes, it currently cannot work if that node is offline. Note that the fallback behavior to other nodes is a bit of bare bones for now, this is definitively something we would like to improve sooner than later.
Yes, just select one of the other nodes in the resource tree in the left panel.
Yeah, something like that is planned.
Not sure what you mean here, PDM is intended to use its own account and permission system.
There is a rudimentary ACL system in the backend already, requires CLI to manage as of now but is planned to be exposed in the UI too.
For a remote? Whatever info and controls you want to expose to PDM. But in general it's intended to give out a very powerful API token for PDM and then have some simple ACL system to control which PDM user can access which Remotes and what they can do with it (think Audit, Modify, ...).
Nothing final yet, but the current plan is for an active-standby system where on PDM is actively querying a set of remotes that is then synced directly to one or more stand-by PDMs that can be promoted to active if the primary PDM is gone.
But such an integrated active-standby system is definitively something for later. The PDM is quite simple and one just setup two and they won't interfere with each other. As of now, one could also basically just rsync
/etc/proxmox-datacenter-manager to another instance.Dragoon
Renowned Member
Nice that proxmox team finally came out with this. But this software has been around for a while and manages the different nodes well.
https://corsinvest.it/cv4pve-admin-proxmox/ https://github.com/Corsinvest/cv4pve-admin
Add the functions and features from them and build it with the proxmox team.
https://corsinvest.it/cv4pve-admin-proxmox/ https://github.com/Corsinvest/cv4pve-admin
Add the functions and features from them and build it with the proxmox team.
I tried that for a while. Only feature request from c4pve i have is to perform and schedule the saving of important files in case of node fault at a central location in PDM.
This will be a great product! And makes managing multiple nodes/clusters much easier!
I've tried the Alpha version, and found some bugs:
1. Sometimes when logging in, the login window will show a infinite "Loading" message, nothing happens. When i refresh the page, i am logged in.
2. Migrate a VM between a cluster and a stand-alone node will show a "api error (status = 400 Bad Request): api error (status = 501 Not Implemented):" message. Both the cluster and node are added via the "root" user to PDM, so it should have the right permissions?
3. When migrating a VM between 2 remote's, the VM ID is not incremented automatically.
I am looking forward to a production version where also PBS is included, so i can have a single control pane for our entire production environment!
I've tried the Alpha version, and found some bugs:
1. Sometimes when logging in, the login window will show a infinite "Loading" message, nothing happens. When i refresh the page, i am logged in.
2. Migrate a VM between a cluster and a stand-alone node will show a "api error (status = 400 Bad Request): api error (status = 501 Not Implemented):" message. Both the cluster and node are added via the "root" user to PDM, so it should have the right permissions?
3. When migrating a VM between 2 remote's, the VM ID is not incremented automatically.
I am looking forward to a production version where also PBS is included, so i can have a single control pane for our entire production environment!
Last edited:
Yeah, we now finally got around to tracking that down and should be fixed with the next version 0.1.9 of the
proxmox-datacenter-manager package. Note that the initial upgrade to that version might still be broken, but once on runs the fixed version upgrades from within the API/UI should run fine and do not interrupt anymore.
There is also a per-node and per-guest open link icon-button down in the resource tree.
