From owner-freebsd-hackers Fri Apr 21 10:13:39 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id KAA11542 for hackers-outgoing; Fri, 21 Apr 1995 10:13:39 -0700 Received: from brasil.moneng.mei.com (brasil.moneng.mei.com [151.186.20.4]) by freefall.cdrom.com (8.6.10/8.6.6) with SMTP id KAA11533 for ; Fri, 21 Apr 1995 10:13:33 -0700 Received: by brasil.moneng.mei.com (4.1/SMI-4.1) id AA12863; Fri, 21 Apr 95 12:12:38 CDT From: Joe Greco Message-Id: <9504211712.AA12863@brasil.moneng.mei.com> Subject: Re: [DEVFS] your opinions sought! To: bde@zeta.org.au (Bruce Evans) Date: Fri, 21 Apr 1995 12:12:37 -0500 (CDT) Cc: bde@zeta.org.au, hackers@freefall.cdrom.com, pritc003@maroon.tc.umn.edu, rgrimes@gndrsh.aac.dev.com, terry@cs.weber.edu, wilko@yedi.iaf.nl In-Reply-To: <199504211627.CAA26034@godzilla.zeta.org.au> from "Bruce Evans" at Apr 22, 95 02:27:48 am X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 2057 Sender: hackers-owner@FreeBSD.org Precedence: bulk > >"ddb: No debugger in kernel" > > >Those of us running unattended production systems cannot afford to have a > >machine crash into ddb, and not reboot. Otherwise, I'd agree... but right > >now, I have to drive a half hour to go reboot systems that lock. > > This is easy to fix. E.g., replace Debugger() by a dummy version and put > the old version in the keyboard driver alone, and maybe disable breakpoint > and trace traps unless the debugger has been reached via the keyboard trap. ...which I wouldn't bother with because my systems will begin running kbdless in the near future (i.e. 2.1R). :-) > >(by the way, this is one of the things that I hope serial consoles will buy > >for me, but haven't gotten around to checking on it. Is there any > >Ctrl-SysRq equivalent available on a serial console? BREAK?) > > BREAK, if it is enabled by a million ifdefs. Serial consoles have the > same problem with automatic reboots. Yes, but then when my pager/etc go off to warn me that a system is down, all I need to do is to telnet/dial in to the box that is holding the other end of the serial console, resume a "screen" session, and switch to the appropriate kermit session. :-) Driving 20 minutes at 3A.M. is enough of a pain in the a** that I can trivially justify the small sum of money it costs me to do something like this. Spending 5 minutes to dial in is better than the 45 minutes currently required for several of the systems. And a single serial card is invariably cheaper than the combination of a keyboard and VGA card, even if it's one of the $32 16550 cards I'm so fond of... I'm not sure just how well this works into the original topic of SCSI-temporary-halts. If Mr. Dufalt(? I think) can come up with a utility to provide similar functionality, that seems like an easier (and maybe more correct?) solution. ... Joe ------------------------------------------------------------------------------- Joe Greco - Systems Administrator jgreco@ns.sol.net Solaria Public Access UNIX - Milwaukee, WI 414/342-4847