[PVE-User] Bonding and packet loss

[IMMO WETZEL]

Could you please provide a network diagram.
Still valid is that lacp is just for a lag(l2 bonding) link between to l2 devices. If your switches are stacked they should be seen as one l2 device. if not, you cannot bond and use lacp. You need IP multipath to use the whole bandwidth of both links. 

Immo

-----Original Message-----
From: pve-user [mailto:pve-user-bounces at pve.proxmox.com] On Behalf Of Daniel
Sent: Wednesday, September 06, 2017 1:27 PM
To: PVE User List
Subject: Re: [PVE-User] Bonding and packetloss

LACP is used for both switches.

My Proxmox Servers are using bonding mode 6 but I get strange bandwith problems:

target host .... host20
------------------------------------------------------------------
 run 1: 	 42.3 Mbits/sec
 run 2: 	 880 Mbits/sec
 run 3: 	 105 Mbits/sec
 run 4: 	 35.9 Mbits/sec
 run 5: 	 36.1 Mbits/sec
------------------------------------------------------------------
 average ....... 219.86 Mbits/sec


-- 
Grüsse
 
Daniel

Am 04.09.17, 15:31 schrieb "pve-user im Auftrag von Mark Schouten" <pve-user-bounces at pve.proxmox.com im Auftrag von mark at tuxis.nl>:

    You cannot just LACP over different switches. It should be a stack of switches.
    
    
    Met vriendelijke groeten,
    
    -- 
    Kerio Operator in de Cloud? https://www.kerioindecloud.nl/
    Mark Schouten  | Tuxis Internet Engineering
    KvK: 61527076 | http://www.tuxis.nl/
    T: 0318 200208 | info at tuxis.nl
    
    
    
     Van:   Daniel <daniel at linux-nerd.de> 
     Aan:   PVE User List <pve-user at pve.proxmox.com> 
     Verzonden:   1-9-2017 22:22 
     Onderwerp:   [PVE-User] Bonding and packetloss 
    
    Hi there, 
     
    here is a small overview if my Network: 
     
    2x HP Switches. Both are connected with 4x 1Gbit with a LACP Trunk to each other – Working as expected. 
     
    Now my Problem, I configured all my hosts with Bond Mode 6 and conncted 1 NIC to Switch One and the other to Switch Two 
    Sometimes I got packetloss and see a Kernel error like this: vmbr0: received packet on bond0 with own address as source address (addr:0c:c4:7a:aa:5c:e4, vlan:0) 
     
    Some hosts are working pretty well and some has Packet Loss. 
    After adding some “rules” to a host which has loss the error messages disappear but the loss (less loss) still exists. 
    Is there any special hint what can be the matter? When I change to active/passive mode all is fine. 
     
    This is my interfaces config which has packetloss: 
     
    auto lo 
    iface lo inet loopback 
     
    iface eno1 inet manual 
     
    iface eno2 inet manual 
     
    auto bond0 
    iface bond0 inet manual 
                    slaves eno1 eno2 
                    bond_miimon 100 
                    bond_mode 6 
     
    auto vmbr0 
    iface vmbr0 inet static 
                    address  10.0.2.111 
                    netmask  255.255.255.0 
                    gateway  10.0.2.1 
                    bridge_ports bond0 
                    bridge_stp off 
                    bridge_fd 0 
                    bridge_maxage 0 
                    bridge_ageing 0 
                    bridge_maxwait 0 
     
    I am absolutely without any glue ☹ tested a lot and nothing really helps to solve this problem. 
     
    -- 
    Grüsse 
     
    Daniel 
    _______________________________________________
    pve-user mailing list
    pve-user at pve.proxmox.com
    https://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
    _______________________________________________
    pve-user mailing list
    pve-user at pve.proxmox.com
    https://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
    

_______________________________________________
pve-user mailing list
pve-user at pve.proxmox.com
https://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user