Hi,
I have a eight node cluster which is all up running and working a treat but when I run the Cluster Validation I receive this warning for each host. I have also included the top few lines of the tests. This warning is repeated from each node to every other node in the cluster.
Warning:
Node HYP01N02.domain.uk is reachable from node HYP01N01.domain.uk by multiple communication paths, but one or more of these paths experienced more than 10% packet loss.
| Result | Source Interface Name | Source IP Address | Destination Interface Name | Destination IP Address | Same Cluster Network | Packet Loss (%) |
| Success | HYP01N01.domain.uk - iSCSI1 | 10.100.1.11 | HYP01N02.domain.uk - iSCSI1 | 10.100.1.12 | True | 0 |
| Success | HYP01N01.domain.uk - iSCSI1 | 10.100.1.11 | HYP01N02.domain.uk - vEthernet (Management) | 172.21.59.12 | False | 96 |
| Failure | HYP01N01.domain.uk - iSCSI1 | 10.100.1.11 | HYP01N02.domain.uk - iSCSI2 | 10.100.2.12 | False | 100 |
| Failure | HYP01N01.domain.uk - iSCSI1 | 10.100.1.11 | HYP01N02.domain.uk - vEthernet (Migration) | 172.21.59.131 | False | 100 |
| Failure | HYP01N01.domain.uk - iSCSI1 | 10.100.1.11 | HYP01N02.domain.uk - vEthernet (Cluster) | 172.21.61.12 | False | 100 |
| Success | HYP01N01.domain.uk - iSCSI2 | 10.100.2.11 | HYP01N02.domain.uk- iSCSI2 | 10.100.2.12 | True | 0 |
| Success | HYP01N01.domain.uk - iSCSI2 | 10.100.2.11 | HYP01N02.domain.uk - vEthernet (Management) | 172.21.59.12 | False | 96 |
Each node has:
iSCSI1 VLAN101 is a 1x 10Gb port and iSCSI2 VLAN 116 is a 1x 10Gb port configured with MPIO
Trunk1 (4x 1Gb NIC) (Address Hash, LACP) on which sits a team with vEthernet (Management) VLAN 101 and vEthernet (Cluster) VLAN 104
Trunk10 (2x 10Gb NIC) (Hyper-V Port, LACP) on which sits a second team with VM traffic and vEthernet (Migration) VLAN 105
Trunk1 and Trunk10 are configured with LACP and locked down to the relevant VLANs on the switch as well as on the vNICs. (Don't know if its best practise to leave the trunk open to all VLANs if the NICs are configured in Hyper-V with VLANs or have both set like they currently are)
iSCSI1 is on VLAN 101 and ip range of 10.100.1.x
iSCSI2 is on VLAN 116 and ip range of 10.100.2.x
vEthernet (Management) is on VLAN 100 and on ip range 172.21.59.12
vEthernet (Cluster) and vEthernet (Migration) are also on there own VLANs and ip ranges.
iSCSI1 and iSCSI2 only has ip v4 and ip v6 left enabled and DNS registration and NETBIOS disabled. They do have Jumbo frames enabled.
Each network can talk over the relevant NIC or vNIC to the other hosts on the same NIC or vNIC which is correct and not across networks which they shouldn't so (Migration) to (Cluster) etc which is also correct.
(Cluster) and (Migration) only allow cluster communications
(Management) accepts cluster and client communications.
iSCSI1 and iSCSI2 are set to NOT allow any cluster traffic.
Binding Order:
1.Management
2.iSCSI1
3.iSCSI2
4.Migration
5.Cluster
Cluster metric is 900 and Migration 1000.
I still receive the warning below and I assume its because some how when testing, there seems to be a tiny conection across VLANs and ip ranges from iSCSI1 or iSCSI2 to the vNIC (Management) which shouldn't be the case. I assume the error is because of the
highlighted line of the table???
Can anyone suggest anything, it is driving me mad! Things seem to work properly but I need the test to complete correctly to be able to sign the clients environment off.
I'm surprised that there is (even if a little) any chat between iSCSI and vEthernet Management according to the test even though they are on different physicalNICs, IP range an even VLANs. Also, trying to establish if the warning is just that a warning
or something to be concerned about. Obviously with all networks broken up as they are, there will be multiple paths between each node of which many are locked down so they cant talk.
Concern is that although it all works fine and even the results read right, I have never seen the warning before in any implementation even when broken up by VLANs, IP ranges and virtual or physical NICs.