Topic: VM Clock problems return after kernel patch

[mudtoe]

Hi folks:

It looks like the clock problems have returned after I applied the patches that upgraded the kernel to 2.6.9-42.0.2  .    When I was running version 2.6.9-34.0.2 putting clock=pit as a boot parameter corrected the problem; however in the current version that doesn't work.  I've tried all the other fixes I could find, except for recompiling the kernel to lower the timer interrupts from 1000 to 100 (not even sure that the kernel source files are available in the standard SME server distribution).  Here is what I have for my boot parameters now:

title CentOS (2.6.9-42.0.2.EL)
   root (hd0,0)
   kernel /vmlinuz-2.6.9-42.0.2.EL ro root=/dev/main/root clock=pit nosmp noapic nolapic acpi=off


The clock is running at only about 50% normal speed.  The only thing that is keeping it from getting too far behind is that I've got vmware tools installed and have "tools.syncTime = TRUE" in the parameter file.  However, this correction only happens once per VM machine minute, so the clock drifts up to two minutes before it gets fixed.  Unfortunately the version of VMWare I'm running (VMWare server 1.01, under Windows Server 2003 64 bit version) doesn't support the "tools.syncTime.period" parameter, so I can't get vmware tools to correct the clock more often.  

Has anyone running SMEserver in a VM gotten the clock to work correctly with this kernel version?  I also tried the SMP version of this patch level of the kernel, just for the heck of it, and it had the same problem.


mudtoe

[svangool]

I also had problems with the clock (see bugtracker, but not related to VMware), for me helped:

Code: [Select]

clock=pmrtmr

Without the other parameters.

Maybe you could try.

Sjef.

[mudtoe]

I tried that one too, no luck            , but thanks for responding.


mudtoe

[robert11]

http://forums.contribs.org/index.php?topic=34516.0

[mudtoe]

I checked out those suggestions, and I've tried most of them.  I'm running vmware tools, which is the only reason that the clock doesn't fall more than a couple of minutes behind.  The rest of the suggestions had to do with the boot parameters, which I've also already tried.  

I'd be really interested in knowing if anyone else is running this version of the operating system in a VM and is able to have the clock work correctly, without vmware tools correcting it every couple of minutes.


mudtoe

[mudtoe]

I went back to version 2.6.9-34.0.2, and the clock is working just fine again with only clock=pit in the boot parameters.  I guess I'll just have to wait until the next version of the kernel comes out and see if that fixes the problem.



mudtoe

[mudtoe]

Does anyone know if the clock synch problem under VMWare has been solved in version 7.1.2?  I'm still on a far back version of SME Server, because when I upgraded the clock wouldn't stay synchronized regardless of what parameters I used.  


mudtoe

[andy_wismer]

Hi

Just use your VMWare Server host, if that is running a ntp. I run sme and others in sme which does include a ntp service. So i just point my ntp service back to the server, which in turn uses our swiss national ntp server...

I've heard we swiss are known for timekeeping...

It works fine! (I am running 7.1.2 on Host and Guest)

YMMV

Andy Wismer

[bdl]

Does anyone know if the clock synch problem under VMWare has been solved in version 7.1.2?  I'm still on a far back version of SME Server, because when I upgraded the clock wouldn't stay synchronized regardless of what parameters I used.

Hi mudtoe,

I've been running 7.1.1 (with no slowdown) with the following parameters:

Code: [Select]

kernel /vmlinuz-2.6.9-42.0.8.ELsmp ro root=/dev/main/root clock=pmtmr nosmp noapic nolapic

Running on VMware server 1.0.1 on a Win2k host.

Hope this helps,

bdl

edit: I've just tested these settings with SME 7.1.2 under a Win2k Host (VMWare Server 1.0.2) and they work perfectly.