Quoth Keith Cantrell on Wed, Oct 22, 2008 at 11:02:21AM -0500: > [ Oct 22 09:43:40 Method or service exit timed out. Killing contract 58 ] > > The last line just keeps repeating every second and has for over an hour > now. I tried doing an "svcadm disable print/server" to no avail.
This is indiciative of a kernel bug -- processes should die when SIGKILLed. > More data: > > # ctstat -i 58 -v > CTID ZONEID TYPE STATE HOLDER EVENTS QTIME NTIME > 58 0 process owned 7 0 - - > cookie: 0x20 > informative event set: none > critical event set: core signal hwerr empty > fatal event set: none > parameter set: inherit regent > member processes: 23371 > inherited contracts: none > # ps -fp 23371 > UID PID PPID C STIME TTY TIME CMD > root 23371 1 0 - ? 0:00 <defunct> > # preap 23371 > preap: Failed to reap 23371: the only non-defunct ancestor is 'init' > # kill -9 23371 > # ps -fp 23371 > UID PID PPID C STIME TTY TIME CMD > root 23371 1 0 - ? 0:00 <defunct> Hmm, this seems pretty bizarre. I would use mdb -k to investigate what's going wrong. Do you know if your init is functioning properly? Does "ps -fp 1" say init is still running? Have there been any messages about init in dmesg or /var/adm/messages? A gcore of init might be useful. Anyway, I suspect you should gather data for a bug report. Quoth Keith Cantrell on Wed, Oct 22, 2008 at 11:22:10AM -0500: > Does anyone know if I can kill the restarter and start it back up? Will > it regain ownership of the existing contracts? Or, am I better off > rebooting? You should be able to kill svc.startd, and the new one should inherit the old one's contracts. If svc.startd is in the middle of starting or stopping services, then some of those processes might be lost. Quoth Antonello Cruz on Wed, Oct 22, 2008 at 09:23:49AM -0700: > try > > svcadm restart restarter > > before rebooting. I'm pretty sure this won't work; svc.startd should ignore commands on itself. David
