Problem with bond mode balance-rr

[kirelgt]

Hello everyone.

The problem I have is weird, I have a 4 NIC server and divide them into two bond0 and bond1. Bond1 is being used to communicate with a NAS system and bond0 which the one used for VMs is giving problem. I forgot to mention that bond0 belongs to vmbr0. Every time I go into the bond-mode and put as balance-rr, all of the VMs stop responding on their network. Then I have to go back and set the bond-mode to active-backup. Active backup is the only mode this system works.

Please help with this issue.

If possible please which is the best bond-mode that I can use.???
Proxmox version
pve-manager/1.8/6070

 

[screenie]

plz post your /etc/network/interfaces and did you some configuration on the switches? what switches do you have?

 

[kirelgt]

Here is my /etc/network/interfaces


# network interface settings
auto lo
iface lo inet loopback

iface eth0 inet manual

iface eth1 inet manual

iface eth2 inet manual

iface eth3 inet manual

auto bond0
iface bond0 inet manual
slaves eth0 eth1
bond_miimon 100
bond_mode active-backup

auto bond1
iface bond1 inet manual
slaves eth2 eth3
bond_miimon 100
bond_mode balance-rr

auto vmbr0
iface vmbr0 inet static
address 10.10.30.37
netmask 255.255.255.0
gateway 10.10.30.1
bridge_ports bond0
bridge_stp off
bridge_fd 0

auto vmbr1
iface vmbr1 inet static
address 192.168.10.89
netmask 255.255.255.0
bridge_ports bond1
bridge_stp off
bridge_fd 0

 

[kirelgt]

The switch I use is a regular switch.

 

[kirelgt]

The switch I use is a [SIZE=-1]HP ProCurve Switch 2900-24G (J9049A)[/SIZE].

 

[screenie]

network config looks ok;
i had also problems using hp switches 2510g-24 where i found out that in bonding mode balance-rr, balance-alb or 802.3af with 2 or 3 interfaces you cannot have more then 1gbit throughput, and also noticed that sometimes vm's are unreachable;
i tried everything you can configure, updated the software on it - nothing helped - could get out more then one link;
after spending days of troubleshooting i came to the conclusion where something in the switching software or fabric must block the traffic when using multiple links;
i opened a ticket with hp support, where this was wasted time - they have no idea what bonding is and tried to explain me staff where they didn't know the spcifications and also not really knowledge of their own products;

after that 3 weeks of frustration i bought 2 alcatel 6400 switches, connected my servers with bonding mode balance-rr and got 1.92gbit throughput with two links and jumboframes on - worked out of the box;

the hp switches are now sitting under my desk doing nothing;
sure they were cheap, but i got what i paid for - cheap, crappy hardware....

if you have other switches around, give them a try - maybe you are dealing with the same shit i had

 

[kirelgt]

Sorry for taking so long to respond.

Well I have to admit that you may have found your problem, but this problem you are describing do not apply to me because we are using VMWARE ESX on other machines connected to this same switch and we are using Load balancing, which means that they use some kind of balancing mode. This proves that it is not the switch.

What could be the problem it is still a mystery and it is driving me crazy.

If someone could help. I will apreciate it a lot.