2014-10-18 16:10 GMT+02:00 Richard W.M. Jones <rjo...@redhat.com>: > Really this should be a warning, not an error, but at least let's > print out the product version so we have a hope of diagnosing the > problem. > > error: internal error: v2v-vcenter is neither an ESX 3.5, 4.x nor 5.x host > (product version = 0x40007) > > In this instance the error was that you have to use 'vpx://' > instead of 'esx://'.
No, this should not be a warning and it cannot be a warning easily without reworking the URI semantic of the driver. The esx:// and vpx:// URI formats are different. An essential difference is that you have to specify the datacenter name, cluster name (conditional) and server name as the path of a vpx:// URI. All of this information has no meaning for an esx:// URI. The driver already warns you that it is going to ignore the path section of an esx:// URI if it's not empty. Which should give you a hint that there is something wrong with your URI. The driver also complains if the path section is malformed of missing from a vpx:// URI. So in the case you mentioned about accidentally using esx:// instead of vpx:// to connect to a ESX serer through a vCenter server, you already get a warning like this: Ignoring unexpected path '/dc1/srv1' for non-vpx scheme 'esx' from virsh -c esx://v2v-vcenter/dc1/srv1 assuming you provided a proper path in you URI, but just mixed up esx:// and vpx://. Also reporting this product version value doesn't really help here, as this is just some driver internal enum value. I propose a different patch instead: https://www.redhat.com/archives/libvir-list/2014-October/msg00517.html This simplifies and relaxes the internal version checks and as a side effect also improves the error messages. In your case it would now report Expecting 'v2v-vcenter' to be a ESX(i) host but found a vCenter/VPX host -- Matthias Bolte http://photron.blogspot.com -- libvir-list mailing list libvir-list@redhat.com https://www.redhat.com/mailman/listinfo/libvir-list
