Description of problem:
I'm working on getting / on iscsi to work for Fedora 10, this requires an initrd fix, see bug 467379. With this initrd fix, / on iscsi works (as long as NetworkManager is disabled), but the following messages are produced by iscsistart during boot:

add: failed - 17: Netlink Error (errno = File exists)
iscsistart: transport class version 2.0-870. iscsid version 2.0-870
iscsistart: version 2.0-870
iscsistart: Logging into iqn.2005-03.com.max:testdisk
192.168.1.100:3260,1
iscsistart: connection1:0 is operational now
iscsistart: transport class version 2.0-870. iscsid version 2.0-870
iscsistart: Can not bind IPC socket
iscsistart: Could not setup mgmt ipc

iscsistart: Logging into iqn.2005-03.com.max:testdisk
192.168.1.100:3260,1
iscsistart: got read error (-1/104), daemon died?
iscsistart: initiator reported error (18 - could not communicate to
iscsid)
iscsistart: can not connect to iSCSI daemon (111)!
iscsistart: initiator reported error (18 - could not communicate to
iscsid)
iscsistart: Could not stop event_loop

I think the first message is not from iscsistart but from nash, but I included it for completeness.

The cause of the iscsistart errors is that iscsistart is being called twice and twice for the same target. I'll investigate this further and come up with a fix, so I'm assigning this to me. Peter I've put you in the CC to keep you in the loop.

Also I've applied for mkinitrd git access, I noticed katzj manages this, but I would like to have your ok (and I assume Jeremy will ask you) for getting mkinitrd access so that I can directly commit patches like the one in bug 467379.