|
Hi Wensong,
Since this message has been posted to the mailing list, I will
take the liberty to answer some of the issues raised.
>I did read your VS patch for kernel 2.2 a week ago.
>
>There are three things in his patch that don't make me feel good.
>I think I have to talk to you directly.
Thank you for your comments on our kernel 2.2 patch. They are
very much appreciated. We have released this experimental
patch as soon as possible so that we are able to incorporate
any comments that others might have and to allow others to
play with the code.
> 1. You just mentioned that the VS patch for kernel 2.2 is
>heavily based on wensong's VS patch for kernel 2.0, it is not
>enough, and it seems that you are the only author. Although
>you changed many variable names and some data structure
>for kernel 2.2 and merge Matthew's loadable load-balancing
>module, the control flow and request techniques is the same
>as the original one, and you shouldn't strip off my original
>license and should mention Matthew's contribution.
>For example, the original networking code is ported from
>FreeBSD networking code, some were rewritten in the kernel
>2.0, and networking code of kernel 2.2 is completely rewritten,
>but the FreeBSD networking code license still remains in the
>kernel 2.2 code. So does IP Firewalling code and IP tunneling
>code and so on, Rusty doesn't strip off the original authors
>because his ipchain code is used for firewalling in 2.2, Alexy
>completely rewritten ipip.c in 2.2 but the original author is
>still the author. The kind of examples is uncountable.
I am not familiar with details of the code, but as far as I know,
it was impossible to port 2.0.36 code directly to the 2.2.x kernel.
The code needed to be completely rewritten from scratch, while
following the logic of 2.0.36 implementation.
According to my understanding, we have included the following license:
> * Virtual Server support for IP Masquerading
> *
> * This code is GPL.
> * It is Experimental software.
> *
> * Written by Peter Kese (peter.kese@xxxxxx)
> * This code is heavily based on the old IP Port Forwarding & Virtual
> * Server code written by Wensong Zhang.
> *
> * For more information, check http://proxy.iinchina.net/~wensong/ippfvs
> * or send an e-mail to peter.kese@xxxxxx
> *
> *
> * Changelog:
> *
> * 1998 - 1999: Linux 2.0 Virtual Server patch
> * Wensong Zhang <wensong@xxxxxxxxxxxx>
> * 05-May-1999: Port to 2.2, complete rewrite of the code, scheduling modules
> * Peter Kese <peter.kese@xxxxxx>
It seems fairly reasonable to me. If you are not happy with the above
license, please send me text that you are happy with.
> 2. You changed the named of ippfvsadm.c to ipvsadm.c and
>insert the loadable load-balancing option (which I guess is based
>on Matthew's change for ippfvsadm.c). However, you striped all
>the original license off, the ipvsadm.c remains without any
>information about authors and licenses. It's not only a good
>programming habit, but violates the GPL.
I apologize for this mistake. We should have included a license here.
We will fix this in the next release.
> 3. You didn't port many functions of the original patch to kernel
>2.2 without any discussion in the mailing list or with me. So the
>VS patches for kernel 2.0 and for kernel 2.2 are not unified. This
>is not what I want to see. I am worrying that lack of communication
>will split the future development of Linux Virtual Server.
Peter has announced on this mailing list that he is working on the
port to 2.2 and he provided source code as soon as it was stable
enough. He also participated in discussions in the mailing list.
Patch for 2.2 kernel has to be very different, because kernel data
structures have changed radically. This is still an experimental
version, so not all features have been implemented yet.
What else would you like to see? The code is available to everyone
to criticize and make comments. We will be happy to receive any
comments or patches for the code. If you want to use our code as
a basis for Virtual Server on 2.2 kernel, we would be very happy
to contribute our future patches to you to maintain a single version.
>By the way, do you know PHT's announcement about their
>TurboLinux cluster in the c.o.l.a. newsgroup, freshmeat.net
>and linuxtoday.com sites? They didn't mention any information
>that they use the LVS code in the TurboLinux Cluster at their site.
> http://community.turbolinux.com/cluster/
>They didn't show any respect to the works of others. It is a
>kind of bad behavior that a good man will never do.
Acknowledgements are made on our front page and on our acknowledgements
page for the official Cluster Server page. Please check out:
http://www.pht.com
http://www.pht.com/clusterwebserver/acknowledge.html
I agree that it would be appropriate to acknowledge your contribution
also on the page that you mention above. I have asked our Webmaster
to fix that. I hope that this will happen soon - we are moving to a
new location, so it might take a day or two.
Thank you,
Rok
--
Rok Sosic
TurboLinux cluster development
rok@xxxxxxxxxxxxxx
|
