Hi,
What's the current status of NFS over IPv6 in Linux? There are HOWTOs out
there, but most patches seem to be against old kernel versions, and
completely unable to run both client and server from the same kernel tree.
Additionally, I see mentions of replacements for portmap, but no clear
upstream URL or updated versions anywhere. Also, there seems to be a separate
git tree maintained by Chuck Lever, but gitweb hides any special patches it
might have in tons of upstream patches.
I'd very much like to be able to sew all of this together at some point,
but it's very hard to get a decent overview of it all. Does anyone have a
"current status" or "plans for the near feature" summary?
/* Steinar */
--
Homepage: http://www.sesse.net/
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
On Fri, Jun 29, 2007 at 06:01:55PM -0400, Chuck Lever wrote:
> I have some patches that add IPv6 support in the RPC client. The server
> is more complicated (how to deal with address-based export rules, and so
> on). There hasn't been much progress on the server side, but I think
> Bull is still working on that.
OK, and they still conflict?
> [snip other useful info, thanks]
> While there is some interest in IPv6 among some of Linux's corporate
> sponsors, there really isn't an overwhelming "killer" requirement for
> IPv6, so there isn't much motivation to get all this completed. I have
> to admit I'm pretty burned out on all this because of how complicated it
> has all become. I'd be happy if someone stepped forward and volunteered
> to help or take over the integration.
Well, all I can really provide is a useful test bed inside Debian. I doubt I
can get the kernel team to include NFS/IPv6-specific patches, but when it
comes to rpcbind, nfs-utils and the like, I'm sure we can work something out.
IOW, I don't think I can do much about the programming, but once stuff is
"80% there" we can at least provide an environment users can test stuff
relatively easy inside. (We already have quite good NFSv4 support
out-of-the-box.)
/* Steinar */
--
Homepage: http://www.sesse.net/
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
Le Rouzic wrote:
>Chuck Lever a ?crit :
>
>> Le Rouzic wrote:
>
>>>>> While there is some interest in IPv6 among some of Linux's
>>>>> corporate sponsors, there really isn't an overwhelming "killer"
>>>>> requirement for IPv6, so there isn't much motivation to get all
>>>>> this completed. I have to admit I'm pretty burned out on all this
>>>>> because of how complicated it has all become. I'd be happy if
>>>>> someone stepped forward and volunteered to help or take over the
>>>>> integration.
>>>>>
>>>>
>>> Some basic tests have been also done on the submitted Aur?lien's
patchs
>>> and we will be running the robustness tests we are regularly doing
>>> for ipv4.
>>> See also at http://nfsv4.bullopensource.org/doc/rpc_testsuite.php
>>> for this list of tests (cthon04, ffsb, fsx...)
>>>
>>>>
>>>> We hope the interest of community about IPv6 will grow.
>>>> Feel free to send comments on the code.
>>>>
>>>>
>>> Some customers are already asking it but were waiting of its
>>> availibity.
>>> There are also products based on NFSV4 features which are interested
in.
>>
>>
>> Yeah, it's kind of unfortunate that we can't be completely specific
>> about the requests/requirements/use cases.
>
>
> Hi,
> I think you are right but nevertheless customers often require to be
compliant IPV6
> in their system even they don't use IPV6. Soon, IPV6 deployed, NFS
will be
> one of the first applications used. Some others UNIX systems also
have it,
> so interoperabity it would be good Linux have it.
> But that is true it will be interested to know when and who on the
nfs lists is interested in
> to use it.
>
>> But here's what I know:
>>
>> 1. Apparently there is a bureaucratic (ie non-technical) requirement
>> from the US Federal government for IPv6 support in new computer
>> software deployments. I forget when this requirement kicks in, or
>> what exactly "IPv6 support" really entails.
Hi,
Back in 2005, the US Office of Management and Budget (OMB) issued a
mandate calling on all Federal Agencies to have an operational IPv6
backbone network by June, 2008. As the date approaches, USG customers are
increasingly asking whether individual products/applications are "IPv6
compliant".
The DoD has also indicated that it is developing an Approved Products List
(APL) on which certified IPv6 compliant products will be listed. The
intention is that any product purchased by the DoD must first be certified
and added to the list.
We have some nongovernment customers (names cannot be disclosed)
requesting for IPv6 support in NFS, and few of them are willing to
participate in pilot program. In addition to the customers, we have some
requests from different product owners for this support, so that their
products can be made "IPv6 complaint".
>>
>> 2. The APAC market wants IPv6 because of an explosion of individual
>> IP network accessors, although arguably that can be addressed for a
>> little while with judicious application of NAT routers.
>>
>> Otherwise, the companies I've worked for recently have had no real
>> customer requests for IPv6 support in NFS.
>>
>> NFS community resources are focused largely on NFSv4.1 and pNFS design
>> and implementation. However, here's some of what I think we need to
>> get IPv6 done:
>>
>> + A coherent design for handling IPv6 address caching and export rules
on the server side
>
> After a first rejected solution and some new discussion on the NFSV4
mailing list,
> Bull (Aur?lien) supplied a new document described the current
> implementation. This little design document is still here:
> http://nfsv4.bullopensource.org/doc/nfs_ipv6.php
>
> Any new remarks are welcome.
>
>>
>> + Help from the netdev folks with reviewing our proposed IPv6 changes
>
> What would be the mailing lists to call them to review?
>
>>
>> + A cunning plan to stage the integration of all the various pieces
>>
>> + Beta testers (ie real environments that can deploy NFS over IPv6 in
real conditions)
>
> May be some on this nfs mailing list have this kind of
environmments.
> What Bull can do is to run the robustness tests we are doing for
IPV4 on IPV6.
>
>>
>> + A whiff of project planning to help us know what the real deadlines
are
>
Thanks.
Regards,
Prasad Potluri
email: pvp at-sign us dot ibm dot com
Chuck Lever a =E9crit :
> Le Rouzic wrote:
>
>>>> While there is some interest in IPv6 among some of Linux's =
>>>> corporate sponsors, there really isn't an overwhelming "killer" =
>>>> requirement for IPv6, so there isn't much motivation to get all =
>>>> this completed. I have to admit I'm pretty burned out on all this =
>>>> because of how complicated it has all become. I'd be happy if =
>>>> someone stepped forward and volunteered to help or take over the =
>>>> integration.
>>>> =
>>>
>> Some basic tests have been also done on the submitted Aur=E9lien's patchs
>> and we will be running the robustness tests we are regularly doing =
>> for ipv4.
>> See also at http://nfsv4.bullopensource.org/doc/rpc_testsuite.php
>> for this list of tests (cthon04, ffsb, fsx...)
>>
>>>
>>> We hope the interest of community about IPv6 will grow.
>>> Feel free to send comments on the code.
>>> =
>>>
>> Some customers are already asking it but were waiting of its =
>> availibity.
>> There are also products based on NFSV4 features which are interested in.
>
>
> Yeah, it's kind of unfortunate that we can't be completely specific =
> about the requests/requirements/use cases.
Hi,
I think you are right but nevertheless customers often require to be =
compliant IPV6
in their system even they don't use IPV6. Soon, IPV6 deployed, NFS =
will be
one of the first applications used. Some others UNIX systems also =
have it,
so interoperabity it would be good Linux have it.
But that is true it will be interested to know when and who on the =
nfs lists is interested in
to use it.
> But here's what I know:
>
> 1. Apparently there is a bureaucratic (ie non-technical) requirement =
> from the US Federal government for IPv6 support in new computer =
> software deployments. I forget when this requirement kicks in, or =
> what exactly "IPv6 support" really entails.
>
> 2. The APAC market wants IPv6 because of an explosion of individual =
> IP network accessors, although arguably that can be addressed for a =
> little while with judicious application of NAT routers.
>
> Otherwise, the companies I've worked for recently have had no real =
> customer requests for IPv6 support in NFS.
>
> NFS community resources are focused largely on NFSv4.1 and pNFS design =
> and implementation. However, here's some of what I think we need to =
> get IPv6 done:
>
> + A coherent design for handling IPv6 address caching and export =
> rules on the server side
After a first rejected solution and some new discussion on the =
NFSV4 mailing list,
Bull (Aur=E9lien) supplied a new document described the current =
implementation. This little design document is still here: =
http://nfsv4.bullopensource.org/doc/nfs_ipv6.php
Any new remarks are welcome.
>
> + Help from the netdev folks with reviewing our proposed IPv6 changes
What would be the mailing lists to call them to review?
>
> + A cunning plan to stage the integration of all the various pieces
>
> + Beta testers (ie real environments that can deploy NFS over IPv6 in =
> real conditions)
May be some on this nfs mailing list have this kind of environmments.
What Bull can do is to run the robustness tests we are doing for =
IPV4 on IPV6.
>
> + A whiff of project planning to help us know what the real deadlines ar
> e
Cheers
-- =
-----------------------------------------------------------------
Company : Bull, Architect of an Open World TM (http://www.bull.com)
Name : Aime Le Rouzic =
Mail : Bull - BP 208 - 38432 Echirolles Cedex - France
E-Mail : [email protected]
Phone : 33 (4) 76.29.75.51
Fax : 33 (4) 76.29.75.18
----------------------------------------------------------------- =
-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
Chuck Lever wrote:
>>
>> I'd very much like to be able to sew all of this together at some point,
>> but it's very hard to get a decent overview of it all. Does anyone
>> have a
>> "current status" or "plans for the near feature" summary?
>
>
> I have some patches that add IPv6 support in the RPC client. The
> server is more complicated (how to deal with address-based export
> rules, and so on). There hasn't been much progress on the server
> side, but I think Bull is still working on that.
The latest server patches are on our web page. I have upgraded it.
http://nfsv4.bullopensource.org/patches/ipv6-server/IPv6_patchset.php
It includes missing parts of code. I have started from a 2.6.21 kernel +
Chuck patchset.
I have send them to Bruce few weeks ago. Bruce did you received the URL
of new patches ?
> Steve Dickson has added Bull's rpcbind (replacement for portmap) to
> Fedora 7, but there is some disagreement about whether to use a
> complete replacement, like rpcbind, or whether to add the needed
> support to the existing portmap code, in general.
>
> Bull has announced a test plan for user-space RPC libraries, although
> I haven't had a moment to look through it. Bull should be providing
> testing resources once we have all of this integrated.
>
A student is developing a RPC/TI-RPC test suite. TI-RPC and RPCbind
adoption by the community has been discussed on the mailing list few
weeks ago.
> While there is some interest in IPv6 among some of Linux's corporate
> sponsors, there really isn't an overwhelming "killer" requirement for
> IPv6, so there isn't much motivation to get all this completed. I
> have to admit I'm pretty burned out on all this because of how
> complicated it has all become. I'd be happy if someone stepped
> forward and volunteered to help or take over the integration.
We hope the interest of community about IPv6 will grow.
Feel free to send comments on the code.
>-------------------------------------------------------------------------
>This SF.net email is sponsored by DB2 Express
>Download DB2 Express C - the FREE version of DB2 express and take
>control of your XML. No limits. Just data. Click to get it now.
>http://sourceforge.net/powerbar/db2/
>
>------------------------------------------------------------------------
>
>_______________________________________________
>NFS maillist - [email protected]
>https://lists.sourceforge.net/lists/listinfo/nfs
>
>
--
********************************
Aurelien Charbon
Linux NFSv4 team
Bull SAS
Echirolles - France
http://nfsv4.bullopensource.org/
********************************
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
Aur=E9lien Charbon a =E9crit :
>Chuck Lever wrote:
>
> =
>
>>>I'd very much like to be able to sew all of this together at some point,
>>>but it's very hard to get a decent overview of it all. Does anyone =
>>>have a
>>>"current status" or "plans for the near feature" summary?
>>> =
>>>
>>I have some patches that add IPv6 support in the RPC client. The =
>>server is more complicated (how to deal with address-based export =
>>rules, and so on). There hasn't been much progress on the server =
>>side, but I think Bull is still working on that.
>> =
>>
>
>The latest server patches are on our web page. I have upgraded it.
>http://nfsv4.bullopensource.org/patches/ipv6-server/IPv6_patchset.php
>
>It includes missing parts of code. I have started from a 2.6.21 kernel + =
>Chuck patchset.
>I have send them to Bruce few weeks ago. Bruce did you received the URL =
>of new patches ?
>
> =
>
>>Steve Dickson has added Bull's rpcbind (replacement for portmap) to =
>>Fedora 7, but there is some disagreement about whether to use a =
>>complete replacement, like rpcbind, or whether to add the needed =
>>support to the existing portmap code, in general.
>>
>>Bull has announced a test plan for user-space RPC libraries, although =
>>I haven't had a moment to look through it. Bull should be providing =
>>testing resources once we have all of this integrated.
>>
>> =
>>
>A student is developing a RPC/TI-RPC test suite. TI-RPC and RPCbind =
>adoption by the community has been discussed on the mailing list few =
>weeks ago.
>
> =
>
You can visit our project web page for more information and find design =
documents at http://nfsv4.bullopensource.org/doc/rpc_testsuite.php
>>While there is some interest in IPv6 among some of Linux's corporate =
>>sponsors, there really isn't an overwhelming "killer" requirement for =
>>IPv6, so there isn't much motivation to get all this completed. I =
>>have to admit I'm pretty burned out on all this because of how =
>>complicated it has all become. I'd be happy if someone stepped =
>>forward and volunteered to help or take over the integration.
>> =
>>
Some basic tests have been also done on the submitted Aur=E9lien's patchs
and we will be running the robustness tests we are regularly doing for ipv4.
See also at http://nfsv4.bullopensource.org/doc/rpc_testsuite.php
for this list of tests (cthon04, ffsb, fsx...)
>
>We hope the interest of community about IPv6 will grow.
>Feel free to send comments on the code.
> =
>
Some customers are already asking it but were waiting of its availibity.
There are also products based on NFSV4 features which are interested in.
Cheers
> =
>
>>-------------------------------------------------------------------------
>>This SF.net email is sponsored by DB2 Express
>>Download DB2 Express C - the FREE version of DB2 express and take
>>control of your XML. No limits. Just data. Click to get it now.
>>http://sourceforge.net/powerbar/db2/
>>
>>------------------------------------------------------------------------
>>
>>_______________________________________________
>>NFS maillist - [email protected]
>>https://lists.sourceforge.net/lists/listinfo/nfs
>> =
>>
>> =
>>
>
>
> =
>
-- =
-----------------------------------------------------------------
Company : Bull, Architect of an Open World TM (http://www.bull.com)
Name : Aime Le Rouzic =
Mail : Bull - BP 208 - 38432 Echirolles Cedex - France
E-Mail : [email protected]
Phone : 33 (4) 76.29.75.51
Fax : 33 (4) 76.29.75.18
----------------------------------------------------------------- =
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs