Sure. This was my mistake! Beside that I think that I experienced two shortcomings in the design of SMF:
1. nonpersist.db corruption by a single disk failure should not happen; keeping the data in the physical memory would be fine. 2. svc should be able to "recover" (reinitialize) after a corruption of its state without requiring a reboot of the maschine's OS. svc.startd should really become a little bit smarter. Anyway I survived the trouble by rebooting the server. Thanks, HJ This message posted from opensolaris.org
