Re: [CentOS] Centosplus vmware kernels....???

2008-05-30 Thread Tru Huynh
On Thu, May 29, 2008 at 05:19:44PM -0500, Tom Bishop wrote:
 Thanks Tru and Johnny, one more question.  Can I just use the
 centos5-testing repo, ie, yum enablerepoxxx install kernel-vm?  I ask
 because I tried and while it worked it loaded an older kernel.  Should I
 just go to tru's directory and install the RPM directly?  Also, ok more than
 one question, are the open-vm-tools in the same repo or only found in
 Johnny's testing directory.  Thanks again...

Hi Les,

Everything under http://people.centos.org/tru/ is signed for testing and
feedback before it can be built by the CentOS build systems and
enter either the regular testing repository or their final destination.
You can look at it as alpha release ;).

Cheers,

Tru
-- 
Tru Huynh (mirrors, CentOS-3 i386/x86_64 Package Maintenance)
http://pgp.mit.edu:11371/pks/lookup?op=getsearch=0xBEFA581B


pgp8HwnEAaJ0t.pgp
Description: PGP signature
___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos


Re: [CentOS] Centosplus vmware kernels....???

2008-05-29 Thread Tom Bishop
Thanks Tru and Johnny, one more question.  Can I just use the
centos5-testing repo, ie, yum enablerepoxxx install kernel-vm?  I ask
because I tried and while it worked it loaded an older kernel.  Should I
just go to tru's directory and install the RPM directly?  Also, ok more than
one question, are the open-vm-tools in the same repo or only found in
Johnny's testing directory.  Thanks again...

On Tue, May 27, 2008 at 8:59 PM, Les Mikesell [EMAIL PROTECTED] wrote:

 Ruslan Sivak wrote:




 Thanks to Tru, kernel-vm is all up-to-date and you can find it here:

 http://people.centos.org/tru/kernel-vm/

 and yes, using the clocksource=pit option should not be an issue with
 these kernels.

 Akemi


 So if I understand this correctly, one should not be using a stock kernel
 when running inside a vm, but should use the kernel-vm kernel?


 It depends on the host hardware and OS - many combinations have trouble
 servicing the 1000hz guest clock which has to be simulated in software.
  Also, some host systems have variable speed CPUs controlled by power
 managment which throws off the guests:

 http://kb.vmware.com/selfservice/microsites/search.do?language=en_UScmd=displayKCexternalId=1591

 --
  Les Mikesell
[EMAIL PROTECTED]


 ___
 CentOS mailing list
 CentOS@centos.org
 http://lists.centos.org/mailman/listinfo/centos

___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos


Re: [CentOS] Centosplus vmware kernels....???

2008-05-27 Thread Akemi Yagi
On Tue, May 27, 2008 at 1:07 PM, Tom Bishop [EMAIL PROTECTED] wrote:
 OK, so I need to bring up a new vm and was wondering what the state of vm
 kernels for centos. I have read this http://bugs.centos.org/view.php?id=2189
 about the tick divider but it wasn't clear what the best step forward is for
 centos 5.1 was, i usually ran with the clocksource=pit option and it looked
 like that and the divider option caused a problem.  I have in the past
 compiled my own but was wondering what others were now doing, thanks in
 advance 8-)

Thanks to Tru, kernel-vm is all up-to-date and you can find it here:

http://people.centos.org/tru/kernel-vm/

and yes, using the clocksource=pit option should not be an issue with
these kernels.

Akemi
___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos


Re: [CentOS] Centosplus vmware kernels....???

2008-05-27 Thread Ruslan Sivak

Akemi Yagi wrote:

On Tue, May 27, 2008 at 1:07 PM, Tom Bishop [EMAIL PROTECTED] wrote:
  

OK, so I need to bring up a new vm and was wondering what the state of vm
kernels for centos. I have read this http://bugs.centos.org/view.php?id=2189
about the tick divider but it wasn't clear what the best step forward is for
centos 5.1 was, i usually ran with the clocksource=pit option and it looked
like that and the divider option caused a problem.  I have in the past
compiled my own but was wondering what others were now doing, thanks in
advance 8-)



Thanks to Tru, kernel-vm is all up-to-date and you can find it here:

http://people.centos.org/tru/kernel-vm/

and yes, using the clocksource=pit option should not be an issue with
these kernels.

Akemi
  
So if I understand this correctly, one should not be using a stock 
kernel when running inside a vm, but should use the kernel-vm kernel?


Russ
___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos


Re: [CentOS] Centosplus vmware kernels....???

2008-05-27 Thread Akemi Yagi
On Tue, May 27, 2008 at 3:33 PM, Ruslan Sivak [EMAIL PROTECTED] wrote:
 Akemi Yagi wrote:

 On Tue, May 27, 2008 at 1:07 PM, Tom Bishop [EMAIL PROTECTED] wrote:

 OK, so I need to bring up a new vm and was wondering what the state of vm
 kernels for centos. I have read this
 http://bugs.centos.org/view.php?id=2189
 about the tick divider but it wasn't clear what the best step forward is
 for
 centos 5.1 was, i usually ran with the clocksource=pit option and it
 looked
 like that and the divider option caused a problem.  I have in the past
 compiled my own but was wondering what others were now doing, thanks in
 advance 8-)


 Thanks to Tru, kernel-vm is all up-to-date and you can find it here:

 http://people.centos.org/tru/kernel-vm/

 and yes, using the clocksource=pit option should not be an issue with
 these kernels.

 Akemi

 So if I understand this correctly, one should not be using a stock kernel
 when running inside a vm, but should use the kernel-vm kernel?

 Russ

You *can* run the distro kernel inside a vm.  The CentOS bug entry
referred to by the original poster explains in great details why a
kernel with 100hz clock rate (kernel-vm) gives you improved
performance compared to the distro standard kernel (1000hz) and
handles clock drifts better.  Recent distro kernels offer a new kernel
option divider= that lets you reduce the clock rate.  This should
eventually eliminate the need for the kernel-vm.  However, at the
moment, the divider= option has a bug that causes problems when it is
combined with clocksource=pit.  And the timer pit is often used when
the system clock tends to go faster.

Akemi
___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos


Re: [CentOS] Centosplus vmware kernels....???

2008-05-27 Thread Tom Bishop
Thanks much, I have in the past compiled my own but thought I would check.
A big THANKS to whoever provides them...

On Tue, May 27, 2008 at 3:17 PM, Akemi Yagi [EMAIL PROTECTED] wrote:

 On Tue, May 27, 2008 at 1:07 PM, Tom Bishop [EMAIL PROTECTED] wrote:
  OK, so I need to bring up a new vm and was wondering what the state of vm
  kernels for centos. I have read this
 http://bugs.centos.org/view.php?id=2189
  about the tick divider but it wasn't clear what the best step forward is
 for
  centos 5.1 was, i usually ran with the clocksource=pit option and it
 looked
  like that and the divider option caused a problem.  I have in the past
  compiled my own but was wondering what others were now doing, thanks in
  advance 8-)

 Thanks to Tru, kernel-vm is all up-to-date and you can find it here:

 http://people.centos.org/tru/kernel-vm/

 and yes, using the clocksource=pit option should not be an issue with
 these kernels.

 Akemi
 ___
 CentOS mailing list
 CentOS@centos.org
 http://lists.centos.org/mailman/listinfo/centos

___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos


Re: [CentOS] Centosplus vmware kernels....???

2008-05-27 Thread Les Mikesell

Ruslan Sivak wrote:



Thanks to Tru, kernel-vm is all up-to-date and you can find it here:

http://people.centos.org/tru/kernel-vm/

and yes, using the clocksource=pit option should not be an issue with
these kernels.

Akemi
  
So if I understand this correctly, one should not be using a stock 
kernel when running inside a vm, but should use the kernel-vm kernel?


It depends on the host hardware and OS - many combinations have trouble 
servicing the 1000hz guest clock which has to be simulated in software. 
 Also, some host systems have variable speed CPUs controlled by power 
managment which throws off the guests:

http://kb.vmware.com/selfservice/microsites/search.do?language=en_UScmd=displayKCexternalId=1591

--
  Les Mikesell
[EMAIL PROTECTED]

___
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos