Subject: Crashes and load problems: Please help!
To: None <port-sparc@netbsd.org>
From: Bruce Lane <kyrrin@bluefeathertech.com>
List: port-sparc
Date: 03/06/2001 07:28:48
Fellow Sun-worshipers, ;-)

	I've got two issues to present for dissection today. The first is a series
of unexplained crashes on the part of my mail server. The system is a SPARC
LX running 1.5 with Qmail, all on 64 megs RAM and a 1.2 gig drive.

	This box was perfectly stable for at least 2.5 months, 7x24, no reboots.
Now, for whatever reason, it will, at random intervals, suddenly halt and
go back to the 'ok' prompt at the console with these errors (also displayed
on the console):

	Watchdog Reset
	Instruction Access Error
	Type 'help' for more information
	ok

	Watchdog Reset
	Memory Address not Aligned
	ok

	I've noticed that this happens just after ntpd gives a message about a
kernel PLL status change to 41. Is there some bug in ntpd that I didn't
read about? The mail server -is- also designated an NTP master for my domain.

	Anyway, after a 'boot' command at the console, it starts up just fine
(except for stopping to clean up the root filesystem on its way), and it
will then be stable for another few days before it does it all again.

	All of my systems are running ntpd, and they've displayed the same bizarre
symptom.

	The second issue revisits the SPARC 5 I was trying to load NetBSD on last
night. I've tried two different hard drives so far, Seagate 32430N's, that
both came out of other Sun systems (which means they have the Sun
firmware), but the failure is very consistent. 

	Specifically: The initial boot and installation startup go just fine, and
then I start getting all kinds of SCSI timeout errors from the hard drive
when the system tries to install the sets from the CD-ROM drive.

	I'm going to try two things to address this: First, I'm going to try a
couple of different SCSI drives. Second, I'm going to try an FTP-based
install.

	Input on either issue would be much appreciated. If I need to build a
kernel for the mail box with some sort of debug tracing in it, so be it; I
just need to know how.

	Thanks in advance.


-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Bruce Lane, Owner and head honcho, Blue Feather Technologies
http://www.bluefeathertech.com  // E-mail: kyrrin@bluefeathertech.com
Amateur Radio: WD6EOS since Dec. '77 (Extra class as of June-2K)
"I'll get a life when someone demonstrates to me that it would be
superior to what I have now..." (Gym Z. Quirk, aka Taki Kogoma).