On Tue, 28 Aug 2007, Marco Neubauer wrote:
> just got my 5501 a few days back.
> It's running fine, except that the rtc is running much faster than realtime.

Known problem.  Lots of people seem to have it, including me.  Seems
to be 5501-specific, as there have been no reports of it happening on
4[58]01 boards.  Read the archives of this group to see reports of
this, and to see proposed cures.

For me, the easiest fix was to run a NTP daemon (as the machine
usually has network connectivity available).  Note that the clock is
so unstable that the NTP daemon had a hard time stabilizing the clock,
and still occasionally gives up (meaning human sysadmin gets an
e-mail, and has to fix the clock and restart NTP by hand).  I'm sure
that configuring NTP better, and maybe switching from OpenNTP to
standard NTP could help with that, but for now it is working good
enough (for my purposes, others might see it differently).  The other
fix that really helps it to configure your OS to use the interval
timer (the option is something like "PIT", on BSD-style OSes it seems
to go into the kernel options, I can get details of this later).

Note that this leaves a huge problem: While the computer is off
(meaning the OS is not running), the BIOS clock will advance extremely
fast.  Which implies that the next time the OS boots, it starts from a
ridiculously wrong time.  Unfortunately, at least OpenNTP will refuse
to move the clock by a large amount when starting, and the machine
will continue to run with a ridiculously wrong time.  For now, my fix
is to manually set the clock with an rdate command on startup (on
Linux, the moral equivalent might be ntpdate).  This then has the
disadvantage that network connectivity is required when booting, since
the system will hang (or start with the wrong time) otherwise.  The
fix is obvious: Never turn the machine off or reboot it; which is a
good strategy anyhow, because on reboot the disk is not found, so
manual intervention is required there.

It would be good if a BIOS update to the 5501 fixed this.  Just like
the IDE disks that vanish on reboot, this problem is not a total
killer (the machine is fundamentally functioning), but it is a hassle
and causes extra work.  And I hate work.

--
Ralph Becker-Szendy    [EMAIL PROTECTED]               (408)395-1435
735 Sunset Ridge Road; Los Gatos, CA 95033


_______________________________________________
Soekris-tech mailing list
[email protected]
http://lists.soekris.com/mailman/listinfo/soekris-tech

Reply via email to