Hi,
Not sure in the scenario which you present below where is the iscsi initiator driver running the guest OS or the parent.
If the initiator is MS iSCSI software initiator – in case of connection loss the initiator waits for 60S (by default, configurable through registries) and within those 60S makes periodic re-connect attempts. If the connection is restored the IO flow continues. If does not at the end of 60S a bus re-enumeration is triggered and if the connection can not be restored the disks are surprise removed.
The wait times are different in case you are using multipathing (MPIO) on top of iSCSI and have multiple sessions to the same target. The bus rescan happens much quicker (15S) in that case so MPIO layer is aware of lost iSCSI sessions (paths in MPIO terminology).
iSCSI Management and Tuning
http://download.microsoft.com/download/D/1/D/D1DD7745-426B-4CC3-A269-ABBBE427C0EF/STO-T738_DDC08.pptx
Best Regards,
Vincent Hu