Hi,
I have a VM with a linux guest configured to use a RAID with VMDK-files on different datastores running on ESX 3.5. In the vmx-file I configured the option scsi0.returnBusyOnNoConnectStatus = "FALSE". When disabling the second datastore (so the second disk of the RAID in the VM guest OS is no longer available), the guest OS continues to work with a degraded RAID and without a hanging system.
If I copy this VM to a vSphere Cluster and remove the second datastore while running the guest OS, the VM hangs as long as the datastore is missing. After reconnecting it, the VM continues to work and the RAID in the guest OS (contrary to the behaviour with ESX 3.5) is not degraded.
For me it seems like the option scsi0.returnBusyOnNoConnectStatus in the vmx-file is no longer working with vSphere. Can anybody confirm this and perhaps also point out, howto to get the function available in ESX 3.5 also in vSphere? I also tried different combinations with the SCSI Controller Type (LSI Logic and BusLogic), but I only got the wanted behaviour after a very large timeout in the guest OS of approximately 1080 seconds.
Hoping for a hint,
Johannes
