Hello Daniel, > I might be missing something, but this database doesn't appear to be > persistent at all - it just lives for the duration of the virConnectPtr > object's lifetime. So if you create two connections you'd get two > different UUIDs for the same VM. Yes, that's right. I thought I could make it work kinda stable before make it persistent. And here I have a little question: The way I thought I could make it persistent is just to implement the virDomainDefineXML function. This way, I would make init__uid_db() check if there is some xml file defined, if not just create a new UUID for that lpar. But, in this case, virsh/libvirtd could be run from anywhere, so the xml file would be in other machine and so on. Then I thought I could create the XML in HMC's machine, then I would have a little more control of the UUIDs. What do you think? []'s -- Eduardo Otubo Software Engineer Linux Technology Center IBM Systems & Technology Group Mobile: +55 19 8135 0885 otubo@xxxxxxxxxxxxxxxxxx -- Libvir-list mailing list Libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list