Hardware requirements or recommendations for PDM ?

[MagicFab]

I've tried PDM from a VM when it was introduced, and would like to have a more permanent setup.

What would be the recommendation for hardware to setup PDM ? It seems illogical to run it from LXC or as a VM in ... Proxmox VE

Any suggestions are most welcome.

---

Fabián Rodríguez | Le Goût du Libre Inc. | Montreal, Canada | Mastodon
Proxmox Silver Partner, server and desktop enterprise support in French, English and Spanish

 

[complexplaster27]

Here's my VM settings, honestly super overkill, never seen it go above 1% cpu usage or 10% RAM usage, like could go with way way less if you don't have the capacity

 

[MagicFab]

Here's my VM settings, honestly super overkill, never seen it go above 1% cpu usage or 10% RAM usage, like could go with way way less if you don't have the capacity

View attachment 93396

That's the thing, your VM is managed from PVE which is managed from the VM ?

Logically PDM would run on baremetal, just not sure if I can just slap it on older hardware with proper RAID/etc. precautions. Low CPU seems to indicate lower end hardware would handle it fine. I would still be nice to hace input from Proxmox on this.

---

Fabián Rodríguez | Le Goût du Libre Inc. | Montreal, Canada | Mastodon
Proxmox Silver Partner, server and desktop enterprise support in French, English and Spanish

 

[Atkatla]

Logically PDM would run on baremetal

I am unsure about this topic too. In VMware World, you woud not run the a vcsa on a cluster which is managed by this vcsa, because the cluster sometimes needs stuff from the vcsa to launch VMs (f.e. if you run vDS-environments).

In Proxmox the cluster control seems currently not to depend on PDM, because PDM (currently) only seems to be an support element for better situation awareness and single-point-of-management in multi-cluster-environments. As long as the corosync stuff in the hosts is fine, a Proxmox cluster should be revoverable without PDM running in the beginning.

As far as I see it from descriptions the recovery plan for an all-cluster-host-down-Szenario (VMs on shared storage) could be:
- turn on enough cluster-hosts back on to fullfill quorum requirements
- connect to cluster via one of the hosts, check cluster state and start PDM
-- or does HA start the VM automatically when it was running before the incident?

If you run PDM on baremetal, it is down when this baremetal host fails. If it is running on a proxmox host, and this host fails, the VM can restart immediately on another hosts (as long as the VM is available from your storage network / cluster). In both cases you have failure probability, but in shared storage-cluster-scenario the recovery seems to be much faster possible.

Is this so or do I overlook something?

 

[ness1602]

Usually with NMS or monitoring systems in big support companies, you have one machine outiside of everything(different power,switch and 3g modem usually) so that when anything or everything dies you get notifications etc. If you are maintaining more than 10-20 clusters than it makes sense to have physical PDM for that.