Hi,
I've built a test cluster in my lab. The kit consists of:
2 Hosts (5 X 1GB nics each).
1 X 16 port switch
1 server with 2012R2 (has the iSCSI target installed.
1 VM on another host which acts as a DC/ DNS for the cluster.
on each host server, nic allocation is as follows:
nic1 (USB): Management network (host 1:192.168.2.40, Host2: 192.168.2.41)
nic2&3 (team): VM Access Network (no ip as it's just a virtual switch)
nic4: iSCSI1 (Host1: 192.168.50.10, Host2: 192.168.51.10)
nic5: iSCSI2 (Host1: 192.168.50.20, Host2: 192.168.51.20)
MPIO is installed and configured.
On the iSCSI target, I have 1 nic for the network and 2 nics for iSCSI (192.168.51.30, 192.168.50.30).
All three servers run Windows Server 2012R2 Enterprise and the two hosts have failover clustering and the hyper-v role installed. All the iSCSI traffic is VLAN'd on the same switch (VLAN50).
There is a Quorum disk on the iSCSI target and a CSV for Virtual Hard Disks. The Quorum disk shows on the owner as Drive Q. The CSV drives can only be seen in disk management or server manager>disks. on both hosts
To test I've created 2 HA VMs and live migrated VM's them between both hosts and that works fine.
I've powered down each host and the VMs come up on the other host(For example, VMs are on Host1 and I power it down. If I log onto host 2, they will then power up).
I've tested the teaming by removing 1 nic cable at a time and that's fine as well. I've tested MPIO/ iSCSI by removing 1 cable at a time and everything is ok.
The issue is as follows:
Scenario1: Disk Quorum Owner is Host 2, I have a running VM on host 1. If I unplug the cables from both iSCSI nics on host 1, Host2 goes offline (I can't ping or connect to it, VMs on Host 1 go off, VMs on Host 2 go off, failover clustering manager on host one cannot find the cluster). If I reboot Host 2, everything works fine. The VMs startup on Host1. Not sure why host2 goes offline. Can anyone please advise?
Scenario2: Disk Quorum Owner is Host 2 owner of data CSVs is Disk 2, I have a running VM on host 2. If I unplug the cables from both iSCSI nics on Host 2, Host2 does not go offline. Instad all CSVs and Quorum move to Host1 ( VMs on Host 1 run as normal, VMs on Host 2 run as normal, failover clustering manager on both hosts work as normal and the csvs are shown as mounted on host2). How is this possible/ is this expected behavior?
Thanks very much,
Regards,
HA