On 10/20/2016 07:45 AM, Alexandre DERUMIER wrote:
Not sure it's easy to achieve currently, as we create the migration through an
ssh tunnel on main node hostname ip.
I think we could try to add a patch for this.
maybe define in /etc/pve/datacenter.cfg an option like : network_migration:
192.168.0.0/24
or merge the migration_unsecure option (which is not exposed in gui) like:
migration: network=192.168.0.0/24,unsecure=on|off
I like the latter more! I look a bit into that.
An option would be also a second host entry like 'pvelocalhost' is, e.g.
pvemigration
Phil opened also a bug report here, this will be used to track the issue:
https://bugzilla.proxmox.com/show_bug.cgi?id=1177
Then when migration occur, we detect src and dst ip which are in theses
networks.
With live local storage migration coming , it could be helpful.
----- Mail original -----
De: "Phil Kauffman" <[email protected]>
À: "proxmoxve" <[email protected]>
Envoyé: Mercredi 19 Octobre 2016 18:23:54
Objet: [PVE-User] Dedicated Migration Network?
I am resubmitting this to the user list after a conversation with Thomas
Lamprecht <[email protected]> during the proxtalks conference.
Is setting up a dedicated migration network a configurable?
How can I get this working?
From what I have seen online I found this hack:
https://forum.proxmox.com/threads/how-to-separate-migration-network-dirty-fix.21538/
# diff QemuServer.pm usr_share_perl5_PVE_QemuServer.pm.erb
4295c4295
< $migrate_uri = "tcp:${localip}:${migrate_port}";
---
$migrate_uri = "tcp:192.168.2.162:${migrate_port}";
This hack does not seem to work either.
On an current PVE it won't, no.
Even after specifically setting up my cluster like so:
pvecm create bravo \
-bindnet0_addr 192.168.2.164 \
-ring0_addr bravo-vmnode2-coro.example.com
pvecm add bravo-vmnode2-coro.example.com \
-ring0_addr bravo-vmnode1-coro.example.com
The 192 network is a bonded 10G interface with a vlan.
After creating the cluster it seems that proxmox ignores my request and
does the following:
# cat /etc/pve/.members
{
"nodename": "bravo-vmnode1",
"version": 4,
"cluster": { "name": "bravo", "version": 2, "nodes": 2, "quorate": 1 },
"nodelist": {
"bravo-vmnode1": { "id": 2, "online": 1, "ip": "10.135.164.162"},
"bravo-vmnode2": { "id": 1, "online": 1, "ip": "10.135.164.164"}
}
}
The 10.135 network is only 1G and only supposed to be used to access the
web interface.
My dns is setup like so:
bravo-vmnode1.example.com has address 10.135.164.162
bravo-vmnode1-coro.example.com has address 192.168.2.162
bravo-vmnode1-priv.example.com has address 192.168.1.162
Each network is dedicated to a particular task. The 192 networks are 10G.
Attached is also a sample of the proxmox nodes /etc/network/interfaces file.
Any insight or links to documentation would be very helpful.
Cheers,
Phil
_______________________________________________
pve-user mailing list
[email protected]
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
_______________________________________________
pve-user mailing list
[email protected]
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
_______________________________________________
pve-user mailing list
[email protected]
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
