[pve-devel] question/idea : managing big proxmox cluster (100nodes), get rid of corosync ?

[Alexandre DERUMIER]

>>But shared storage between different clusters is a problem, because our
>>locking mechanism only works inside a cluster. So there must be a single
>>cluster which does all allocation for a specific storage??

But if we have unique id (uuid for example), it shouldn't be a problem ? no need to cross clusters storage lock ?


----- Mail original -----
De: "dietmar" <dietmar at proxmox.com>
À: "aderumier" <aderumier at odiso.com>, "pve-devel" <pve-devel at pve.proxmox.com>
Cc: "Thomas Lamprecht" <t.lamprecht at proxmox.com>
Envoyé: Mardi 20 Septembre 2016 17:37:08
Objet: Re: [pve-devel] question/idea : managing big proxmox cluster (100nodes), get rid of corosync ?

> On September 20, 2016 at 10:45 AM Alexandre DERUMIER <aderumier at odiso.com> 
> wrote: 
> 
> 
> >>I guess we need more. For example, we can exactly identify 
> >>a VM/CT/Storage/Network by prefixing it with the cluster name: 
> >> 
> >>cluster1/vm/100 
> >>cluster2/ct/100 
> >>cluster1/storage/local 
> >>cluster2/network/vmbr 
> >> 
> >>But we need a way to tell if resources are considered 
> >>to be equal (cluster wide). 
> >> 
> >>An easy way to to do that would be to assign UUIDs to VMs, CTs, 
> >>storages and networks? 
> 
> 
> yes, could be the solution. I think openstack/xenserver already doing this. 
> 
> The vm disks also need to be unique. (they use vmid, so if vmid is unique it's 
> ok) 

But shared storage between different clusters is a problem, because our 
locking mechanism only works inside a cluster. So there must be a single 
cluster which does all allocation for a specific storage??