Alan DuBoff wrote: > On Sun, 16 Sep 2007, [EMAIL PROTECTED] wrote: > >> I assume you know about the "snooping" flag and deadman timer. This >> can be >> useful in getting a dump when the system is "hard" hung (i.e., you may >> or may not >> have kmdb loaded, but you can't get in). > > Recentely I've had a system that wouldn't come back even with the > deadman turned on. This can happen if an interrupt is wedged at high enough level to interfere with the real time clock. I have an application that allows you to program the real time clock so that it is doing nothing but firing high resolution timers and the default 50 second deadman timer takes a couple of hours... > >> I think better is a flag that says that when the system panics, bypass >> kmdb and go immediately >> to the dump code. And I think this should be the default behaviour. I >> mostly use kmdb >> for analyzing hangs, not panics. When the system panics, you're >> generally too late anyway, and I would >> rather run mdb on the dump (so I have full access to the machine >> resources). > > Do you mean you so you don't have to type systemdump? Don't you like > to look around before you take a dump? :-/ I think this has been covered in subsequent emails between Joerg and John Levon. I don't feel strongly about the default behaviour. And as for looking around before taking a dump, if I am remote (or it is 2am), maybe I would rather let the machine take the dump for me. It'll be easier to look at on a running system. max
> > Maybe I am misunderstanding you. > > -- > > Alan DuBoff - Solaris x86 IHV/OEM Group > _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org