a tidbit

VMWare rtc: lost some interrupts

You run VMWare Server on Debian Linux. You start a Guest OS, and suddenly your logs start filling up:
Aug 21 12:56:11 dey kernel: rtc: lost some interrupts at 2048Hz.
Aug 21 12:56:42 dey last message repeated 1528 times
Aug 21 12:57:43 dey last message repeated 3050 times
Aug 21 12:58:44 dey last message repeated 3050 times
VMWare is being a little too aggressive about checking the clock.

I'm sure there are better ways to fix this, but the VMWare Timekeeping Manual recommends the following workaround for Guest OSes where exact timekeeping isn't necessary:

You can prevent /dev/rtc from being used. This will generally cause clocks to run slow
in any virtual machines you have that need the additional interrupts, but that may be
acceptable to you, depending on your application. To do so, add the following setting to
each virtual machine's .vmx configuration file, or add the setting globally to the host's
configuration file (/etc/vmware/config):
host.useFastClock = FALSE
Added that line to /etc/vmware/config, called /etc/init.d/vmware restart, and the messages disappeared.

Update: As pointed out here, you can also fix the problem (rather than just ignoring it as I did) by building a custom kernel with HPET_EMULATE_RTC.

I don't use VMWare anymore, but if you do, you should definitely take the time to grok the High-Precision Event Timer and the various kernel options related to it.

By Chris Snyder on August 21, 2006 at 1:29pm

jump to top