[15193] in Athena Bugs
sun4 8.1.4: savecore lossage
daemon@ATHENA.MIT.EDU (John Hawkinson)
Tue Jun 10 20:41:51 1997
Date: Tue, 10 Jun 1997 20:41:43 -0400 (EDT)
To: testers@MIT.EDU
Cc: bugs@MIT.EDU
From: John Hawkinson <jhawk@MIT.EDU>
It is rather inconvenient to enable saving of crash dumps
under Athena Solaris 8.1.4.
/etc/init.d/sysetup contains the following:
##
## Default is to not do a savecore
##
#if [ ! -d /var/crash/`uname -n` ]
#then mkdir -m 0700 -p /var/crash/`uname -n`
#fi
# echo 'checking for crash dump...\c '
#savecore /var/crash/`uname -n`
# echo ''
And in order to enable this those lines must be uncommented.
But sysetup is a tracked file, so those changes are obliterated by
an update.
I used to maintain a .private file that patched sysetup, but this
was a pain, and the exact nature of the patch changed from 8.0 to 8.1,
so this broke.
Some solutions are to:
1. Make /etc/init.d/sysetup a configuration file, and therefore not
tracked.
2. Create a "mkserv savecore"
3. Continue to allow users to bludgeon themselves painfully when they
encounter kernel problems.
4. Add support for running savecore to /etc/init.d/athena, and restore
SAVECORE to /etc/athena/rc.conf. It's not clear to me that this
necessarily happens at the right part of the boot process to
guarantee you a valid core...
5. Consider yelling at Sun for forcing users to edit non-configuration
files to get this stuff. Most scripts in /etc/init.d use some other
method of deciding to do stuff, like various checks for the existance
of files such as /etc/notrouter, or /etc/defaultdomain, etc., etc.
6. ??
I'm kind of at a loss as to what the best solution is. Greg seemed
to be in favor of (2), but I don't really like the idea for somewhat
inexplicable reasons that might have something to do with a general
dislike of mkserv.
--jhawk
