Greetings,

I have been given the responsibility to build a virtual server cluster for the computer science IT department of a small university. The setup uses two Drobo elite units, each with six 2TB drives, and a couple ESXi hosts clustered together with vCenter. Each host has two NICs--one dedicated for iSCSI traffic. Additionally, we have two switches, one for the iSCSI traffic and one for regular traffic.

I've run into an issue on one of my RedHat 6 VM's with iSCSI. I am connecting to a volume on the Drobo unit am able to login via iscsiadm just fine. The problem is, if I try and rsync a tree of files from the iSCSI disk to the virtual hard disk of the VM, it seems to transfer in bursts and then stop for 5-10 seconds at a time. I get connection x: ping timeout errors in /var/log/messages.

At first I thought this might be due to flow control, but I ensured that it is enabled on the switch (HP ProCurve 2910), the ESXi host, and the RedHat VM. I did however notice that I could only get RX frames to enable in RedHat, while on the ESXi host I noticed that both rx and tx frames were enabled. Is that a problem?

Next, I've noticed that when the iSCSI initiator connects, a message is sent to /var/log/messages saying, "sd n:0:0:0: [sdx] Got wrong page" followed by, "sd n:0:0:0: [sdx] Assuming drive cache: write through"

I've looked all over the Internet trying to find out what this means, and nobody seems to know. If you do, could you enlighten me? Could this be a symptom of the problem?

Thanks.