Loss of iSCSI SAN connection causes ESXi 5.0 Hosts to become non-responsive

[virtualG78]

We are having a problem regularly on ESXi servers whereby any time a iSCSI connection is lost, the ESXi host falls into a non-responsive state.  Even after the iSCSI target is confirmed to be connectable again, the ESXi host will not automatically restore the connection.  The only solution we have found is to hard reset the ESXi host as all other methods, such as a soft "reboot" command or a reboot request from vSphere client are non-responsive.  This problem occurs regardless of whether there are running VMs using the iSCSI SANs as storage or not.

The /var/log/messages log on the ESXi hosts show a continuous loop of failed attempts to restore iSCSI access. 

While I realize that iSCSI connections should be reliable, is it possible to restore the iSCSI connection on the ESXi host without needing to resort to a hard reset of the server?

[MasonRupert]

What manufacturer servers are you running? Hardware or software iSCSI? How many paths are you connecting and how many datastores are presented? 

[virtualG78]

ESXi servers are HP DL385 G7s, DL360 G7s, DL360 G6s and Sun Sunfire 4170s.

SANs are HP P4000 and Iomega StorCenter px4-300r.

iSCSI controllers are Broadcom iSCSI controllers on HP systems and Software iSCSI on Sunfires. 

iSCSI target paths vary from 9 - 20 from host to host, one for each used datastore, with up to 7 - 8 target paths on the Iomega SAN (the SAN that loses connection frequently)