On Sat, Dec 16, 2006 at 06:39:09PM +0100, Vclav Haisman
wrote:> Hi, I have loads of following messages in newly installed virtual server
> (under MS Virtual Server 2005 R2) running 6.2 RC1, I am even using the
> stock kernel. Can I fix this?
> Dec 16 18:33:27 shell kernel: calcru: negative runtime of -553620 usec
> for pid 76484 (zsh)
> Dec 16 18:33:27 shell kernel: calcru: runtime went backwards from
> 1004708 usec to 791507 usec for pid 655 (sendmail)
You can reduce them somewhat by using a slower clock tick and perhaps a
different timecounter:
==> /boot/loader.conf <=kern.hz="200"
==> /etc/sysctl.conf <=kern.timecounter.hardware=ACPI-safe
The above reduced my errors to a few hundred a day and improved time
keeping. Setting up the host as an NTP server and running
ntpdate once a minute seems to have sufficed to keep time accurate
enough for my needs.
I got my clues from VMWare documentation <grin>.
Other problems I've seen:
FreeBSD won't even boot if there is a SCSI controller attached to the
guest. I still had my boot drive attached to IDE.
The kernel sometimes (I've seen this twice in a couple of months) loops
outputting a message which might indicate that it disliked the disk
drive and has detached it.
With a NT guest, iozone sometimes (after hours of disk battering)
reports that it read something other than what it had written. I've not
seen iozone do this on the MS 2003 host, on the same disks.
The last two make me fear that the IDE emulation sometimes glitches.
--
Adrian Wontroba
Prof: So the American government went to IBM to come up with a data
encryption standard and they came up with ...
Student: EBCDIC!