[PL #3920] Re: IPv6 support for PlanetLab
hetao at cernet.edu.cn via RT
devel at planet-lab.org
Tue Jan 25 03:57:14 EST 2005
Email Recipients (see http://www.planet-lab.org/Support)
Requestor: hetao at cernet.edu.cn
Ticket Ccs: mef at cs.princeton.edu, mlhuang at cs.princeton.edu
==================================================
Tue Jan 25 03:57:14 2005: Request 3920 was acted upon.
Transaction: Ticket created by hetao at cernet.edu.cn
Subject: Re: IPv6 support for PlanetLab
Hi, Marc and Mark,
We are ready to collaborate with
you two now. So what should we do
at first ? VNET code analysis or
read Mark's documentation about
that ?
|
Regards,
-Tao
======= 2005-01-25 03:50:00 Marc E. Fiuczynski wrote: =======
>Hello Tao,
>
>While we are interested in bringing
>IPv6 support to PlanetLab, we have to
>rely on your support (time and effort)
>to actually do so. We can certainly
>turn on the IPv6 stack, which is a
>simple kernel compile option. But
>there is additional effort required to
>get VNET to do the right thing. VNET
>is our support to expose raw sockets
>and other networking support properly
>to the vservers.
>
>Mark Huang is working on the
>documentation and general clean up of
>the VNET support. Either you or
>someone else in China will need to
>collaborate with Mark to make VNET
>IPv6 capable. Personally, I am not
>sure what effort is involved. It
>might simply be a few lines in the
>kernel that mirror what was needed for
>IPv4. There may also be some issues
>with connection tracking support,
>which is pretty strong for the IPv4
>code base but an unknown (at least to
>us) for the IPv6 code base.
>
>I am cc'ing Mark and am making this a
>devel topic for tracking purposes.
>
>Marc
>
>
>> -----Original Message-----
>> From: Tao HE
>[mailto:hetao at cernet.edu.cn]
>> Sent: Monday, January 24, 2005 9:56
>AM
>> To: Marc E. Fiuczynski
>> Subject: Re: RE: RE: Marc, I'm Tao
>of CERNET
>>
>>
>> Marc,
>>
>> We are very pleased to hear that
>Princeton
>> will bring IPv6 support into new
>PlanetLab OS.
>>
>> I have tried installing Fedora Core
>2 these
>> days and configured it to support
>IPv6. And
>> I also discussed with others to
>collect
>> different ideas.
>>
>> To "enable", we just have to compile
>IPv6
>> support into the kernel or install
>kernel
>> module ipv6.o. In modern Linux
>distributions
>> fundemental tools such as "ifconfig"
>and "
>> route" all support IPv6.
>>
>> As for the daemons/services, we
>would like to
>> run http/ftp/dns/email/ssh services
>to operate
>> IPv6 on CERNET2.
>>
>>
>|
>> Regards,
>> -Tao
>>
>> ======= 2005-01-18 01:18:00 Marc E.
>Fiuczynski wrote: =======
>>
>> >Tao,
>> >
>> >Since we are not IPv6 experts,
>could
>> >you give me an outline of what
>> >software you must run locally to
>> >enable a regular Fedora Core 2 box
>> >with IPv6 functionality (beyond
>> >turning on the IPv6 stack in the
>> >kernel). For example, what
>> >configuration files must be setup
>and
>> >what daemons (if any) must run to
>> >operate IPv6 on CERNET2?
>> >
>> >We are working on bringing native
>IPv6
>> >to Princeton. While most of the
>> >technology for this is in place, we
>> >(PlanetLab @ Princeton) would be
>the
>> >first clients to actually use this
>> >facility.
>> >
>> >In terms of the PlanetLab Linux
>> >kernel, the primary work necessary
>is
>> >in enabling Mark Huang's vnet
>packet
>> >with IPv6 support. This requires
>some
>> >code, but not much. The challenge
>is
>> >more in learning how IPv6 works and
>> >then finding the right places to
>hook
>> >vnet into the existing IPv6 stack.
>> >
>> >So, I think the best way to move
>> >forward is with you providing us
>with
>> >a high-level description of which
>> >configuration files and daemons are
>> >necessary to operate IPv6 on
>CERNET2.
>> >
>> >Best regards,
>> >Marc
>> >
>> >> -----Original Message-----
>> >> From: Tao HE
>> >[mailto:hetao at cernet.edu.cn]
>> >> Sent: Tuesday, January 18, 2005
>1:00
>> >AM
>> >> To: Marc E. Fiuczynski
>> >> Subject: Re: RE: Marc, I'm Tao of
>> >CERNET
>> >>
>> >>
>> >> Marc,
>> >>
>> >> Sorry for the delay. These dyas I
>am
>> >busy with the setup and maintain
>> >> of distributed machines
>throughout
>> >China. The difficulty in
>> >> communication between local tech
>and
>> >I is more than I expected. As
>> >> you may know, I cannot figure out
>> >what happened to the unreachable
>> >> nodes easily if no any
>information
>> >about that. I can neither login in
>> >> the machine nor see the output on
>> >the monitor, so the only way I can
>> >> do is guessing what happened,
>> >telling them to check up, and
>finding
>> >> out how to resolve problem. So,
>it
>> >IS a very long time.
>> >>
>> >> But don't worry about that.
>Things
>> >are better and better since local
>> >> techs are familiar with the
>> >PlanetLab...
>> >>
>> >> Back to our purpose, the next
>step
>> >of PlanetLab in CERNET will be the
>> >> development of IPv6 supported
>> >PlanetLab, which we called
>6PlanetLab.
>> >> So if you can offer some advice
>> >about that, we will be very
>> >appreciated.
>> >>
>> >> Regards,
>> >> Tao
>> >>
>> >> ======= 2004-12-16 17:48:00 Marc
>E.
>> >Fiuczynski wrote: =======
>> >>
>> >> >Hello Tao,
>> >> >
>> >> >Pardon that I had to leave this
>> >> >afternoon without chatting with
>> >you.
>> >> >Please do not apologize for your
>> >lack
>> >> >of being able to express
>yourself.
>> >> >Lets just say that your English
>is
>> >> >infinitely better than my
>Chinese.
>> >> >
>> >> >As mentioned, our Linux kernel
>is
>> >> >based on 2.6 with Fedora Core 2
>> >> >modifications. This is then
>> >enhanced
>> >> >further with CKRM, Vserver,
>VNet,
>> >and
>> >> >a number of other external
>> >packages.
>> >> >Rather than describing the
>overall
>> >> >architecture in detail, it will
>be
>> >> >easier if you could identify on
>> >what
>> >> >parts of the system you or your
>> >peers
>> >> >would consider working on in the
>> >> >future. Are you primarily
>> >considering
>> >> >doing work with the underlying
>> >> >networking facilities? Or are
>you
>> >> >also considering looking into
>> >> >developing new resource
>controllers
>> >> >(e.g., for CPU scheduling,
>memory /
>> >> >disk space, disk io, etc.)?
>> >> >
>> >> >Best regards,
>> >> >Marc
>> >> >
>> >> >
>> >> >> -----Original Message-----
>> >> >> From: Tao HE
>> >> >[mailto:hetao at cernet.edu.cn]
>> >> >> Sent: Thursday, December 16,
>2004
>> >> >3:15 PM
>> >> >> To: mef at cs.princeton.edu
>> >> >> Subject: Marc, I'm Tao of
>CERNET
>> >> >>
>> >> >>
>> >> >> Hi, Marc,
>> >> >>
>> >> >> I always mis-express myself.
>> >> >> My answer to your quesion this
>> >> >afternoon is:
>> >> >> First, we try to understand
>how
>> >it
>> >> >works.
>> >> >> Then, we want to work on it.
>> >> >> That's it.
>> >> >>
>> >> >> Regards,
>> >> >> Tao
>> >> >>
>> >> >>
>> >>
>> >> = = = = = = = = = = = = = = = = =
>=
>> >= =
>> >>
>> >>
>> >
>> >
>> >.
>>
>> = = = = = = = = = = = = = = = = = =
>= =
>>
= = = = = = = = = = = = = = = = = = = =
More information about the Devel-community
mailing list