> We are running RedHat Enterprise Linux 4 (AS) on Microsoft Virtual Server
> R2. The RHEL OS all installs (from RIS via PXE Boot) fine, applications
> and system kernel are patched to the latest revisions. Everything appears
> to operate correctly, all except the time.
>
> The system time when operating under Linux runs too fast - the Virtual
> RHEL system gains something like 5 mins every half an hour. Unfortunately
> this is too much gain for NTP to handle and so the system never gets to
> synchronise with any of our time servers. Manual synchronisation with the
> NTP server (stopping "ntpd" then manually running "ntpdate") works fine,
> however it only takes a few seconds for the system time to be wildly
> inaccurate. Attempts at reducing the clock "ticks" using "tickadj", whilst
> unpleasant, do go some way to working, but the clock still runs too fast.
>
> Our next thoughts of action are to look into kernel parameters/tweaks
> which may adjust the way the system perceives time, or possibly creating
> an "ntp.drift" file with a massive drift value. However this won't help
> our build/patching processes as the same build works fine on a "real"
> system.
>
> Has anyone else come across this behaviour and are there any known
> work-arounds or fixes?
>
> Jim
> --
> Jim Watts
> Information Systems Services
> University of Southampton
>
>

>
>Jim: i have this same problem. also running redhat enterprise as 4
>under ms virtual machine. i tried the suggestions in the vm article.
>with them i can either run fast or slow but cant run close enough for
>ntp to keep me acuate. have you found anything else? thanks.