VMware ESXi 4.1U1. Multipathed iSCSI setup to an IBM MD3000 array.
iSCSI VMkernel IP addresses were set up the same between two hosts. They should have been different.
IPs on host 1: 10.1.1.10 and 10.1.2.10
IPs on host 2: 10.1.1.10 and 10.1.2.10
iSCSI worked 99.999% of the time. Roughly three times an hour, one of the SQL Servers on one of the two hosts would drop a connection or two.
I don’t get how this could have worked from the start… any suggestions?