Quoth Hans-Joerg on Wed, Aug 22, 2007 at 08:57:59AM -0700: > This morning one of the mirrored bootdisks of our nfs-server (a V240) > has been replaced. At that time a core of svc.configd has been created > in /. The running standalone daemons (nfs, ssh, etc) are fully > functional, but svc.configd fails to start. > > db_errors contains this (timestamp matches with the incident): > > /etc/svc/volatile/svc_nonpersist.db: PRAGMA integrity_check; failed. Results: > file is encrypted or is not a database > > And svc_nonpersist.db has changed at this time. I want to avoid > a reboot of the machine very much. So I think restore_repository isn't > suitable. > > Does anybody have a hint for me? Any help is highly appreciated.
That's a tough one. svc_nonpersist.db contains your runtime state (in particular, the state of each service and its contract ID), and there are no backups. I think you could delete it and let svc.configd restart, but I imagine svc.startd would try to start all of your services, which wouldn't be pretty given that all of your old processes are still running. Please send us (me, I suppose) a copy of that core file and your svc_nonpersist.db, and then I think you'll have to reboot the system. Make sure you copy svc_nonpersist.db somewhere persistent first, though, since /etc/svc/volatile is a tmpfs filesystem. David
