Hi,
I am really, really struggling to get networking correct on a Hyper-V cluster. The public LAN (Management and virtual machine network) is fine. This is purely focusing on the network adapters involved in the cluster.
I have 2 cluster nodes, and 4 switches. I have 1 server and 2 switches in each building, the switches are HP 2920.
We will call building 1 ServerRoom, and Building 2, BackupRoom
Switch 1 in ServerRoom is connected to switch 1 in BackupRoom on port A1 of the switch by fibre optic
Switch 2 in Server Room is connected to switch 2 in BackupRoom on port A1 of the switch by fibre optic
Switch 1 and 2 in ServerRoom are stacked
Switch 1 and 2 in BackupRoom are stacked
Below shows the IP addresses on each server and what ports they are connected to on the switches
This is how my cables go from server to the switches
ServerRoom
Switch Network Port IP Address
sw1 iscsi Primary 1/a2 10.10.1.1
sw1 Live Mig Primary 1/1 10.10.2.1
sw1 HB Primary 1/2 10.10.3.1
sw2 iscsi Backup 2/a2 10.10.5.1
sw2 Live Mig Backup 2/1 10.10.6.1
sw2 HB Backup 2/2 10.10.4.1
BackupRoom
Switch Network Port IP Address
sw1 iscsi Primary 1/a2 10.10.1.2
sw1 live mig Primary 1/1 10.10.2.2
sw1 HB Primary 1/2 10.10.3.2
sw2 iscsi Backup 2/a2 10.10.5.2
sw2 Live Mig Backup 2/1 10.10.6.2
sw2 HB Backup 2/2 10.10.4.2
I have created a trunk between switch 1 and switch 2 on the fibre optic ports
Configure
Trunk 1/a1,2/a1 trk1 lacp
I have ran the same command on the switch in BackupRoom
Because I have a primary physical network cable and a backup physical cable in each server for each cluster resource I have to put these in different subnets, so they all use subnet mask 255.255.255.0
I have created 6 vlans, on each switch stack to keep the traffic separate, but I have then tagged trk1 into each vlan so I can benefit from the LACP
These commands are on the switch stack, I have used an IP ending in 5 (10.10.1.5 for example) to designate as the “management IP” or whatever HP refers to it as for each vlan (I am not too good with switches), and in BackupRoom I have used an IP ending in 6 (10.10.3.6 for example)
Vlan 2
Ip address 10.10.1.5 255.255.255.0
untag 1/a2
tag trk1
vlan 3
ip address 10.10.2.5 255.255.255.0
untag 1 /1
tag trk1
vlan 4
ip address 10.10.3.5 255.255.255.0
untag 1/2
tag trk1
vlan 5
ip address 10.10.5.5 255.255.255.0
untag 2/a2
tag trk1
vlan 6
ip address 10.10.6.5 255.255.255.0
untag 2/1
tag trk1
vlan 7
ip address 10.10.3.5 255.255.255.0
untag 2/2
tag trk1
I have repeated the above for BackupRoom switch stack.
However, I use a product called starwind to synchronise the storage, and when I use this configuration it seems to crash the product. When I remove the network cables from one of the switches or remove both the ISCSI cables the product frees itself up. This would indicate there is something wrong with my switch configuration, as if the network is being flooded (by the way I enabled spanning tree protocol too). So I’m really stuck on this….. can anyone suggest anything?
Thank you
Steve