Ben Rockwood wrote: > Serge wrote: [Cc'ing -bugs]
>> Hi all, >> >> Little problem, and strange (Sol10x86 6/06 on a Pentium 4 RAM 512) : now at >> most shutdowns by init 5, the system becomes as well down, but instead to >> make the power off, it reboots after a beautiful memory dump. After that, i >> find the dump files in /var/crash/my_machine as normal. And this dump + >> reboot, it can make its two times, i.e. after a new command "init 5", after >> that it makes really the power off without crash dump. >> >> With init 0, it becomes normally down, and i can switch the power manually >> off. >> >> During the dump, i readed something about PCI, but the whole operation is >> relatively fast. >> >> This issue can happen if before i greatly used the memory with any graphical >> app, but not necessary. >> >> The only PCI device i recently changed is the videocard, my new is a GeForce >> 6600, but it seems since a month properly installed and working. >> >> That all isn't a very big problem, but have anyone similar encourred ? >> > > You can avoid this by disabling "savecore". Just run "dumpadm -n". > > root at aeon ~$ dumpadm -n > Dump content: kernel pages > Dump device: /dev/dsk/c3d0s1 (swap) > Savecore directory: /var/crash/aeon > Savecore enabled: no > If the system is crashing at shutdown, that's a bug. The solution isn't to disable savecore, but to fix the cause of the crash. Could you run: # cd /var/crash/<hostname> # mdb <dump number> # (e.g. vmcore.2 would be 2) > ::status > *panic_thread::findstack > ::stackregs > ::cpuinfo -v And file a CR at bugs.opensolaris.org, including the output from the above? -- Rich