Re: userland nfs server

Re: userland nfs server

[Matt W. Benjamin]

Hi Sage,

We (Linux Box) have built and published a native FSAL for Ceph, which is pretty advanced, and we've submitted the FSAL driver code upstream for review Philippe Deniel.

In addition, we're working on pNFS support in Ganesha and Ceph, and have a prototype file-based pNFS layout driver for the FSAL that takes advantage of Ceph OSDs for parallel IO.

We're actively extending these efforts and there should be some publication forthcoming.

We have an external git repository here:

    git://github.com/aemerson-linuxbox/ganesha.git

We would really appreciate your feedback--and guidance on how to participate in Ceph, we'd like to.

Matt

----- "Sage Weil" <sage@newdream.net> wrote:

> Has anyone looked at Genesha?
> 
> 	http://sourceforge.net/apps/trac/nfs-ganesha/
> 
> In principle it would probably be pretty straightforward to wire that
> up 
> to libceph and avoid the kernel NFS server for reexporting ceph via
> NFS...
> 
> sage
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel"
> in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

-- 

Matt Benjamin

The Linux Box
206 South Fifth Ave. Suite 150
Ann Arbor, MI  48104

http://linuxbox.com

tel. 734-761-4689
fax. 734-769-8938
cel. 734-216-5309

Re: userland nfs server

[Brian Chrisman]

Interesting... I'll pull your Ganesha code and do some testing... I
don't see any docs, should I assume there's little difference with
Ganesha?

On Fri, Jul 29, 2011 at 9:30 AM, Matt W. Benjamin <matt@linuxbox.com> wrote:
> Hi Sage,
>
> We (Linux Box) have built and published a native FSAL for Ceph, which is pretty advanced, and we've submitted the FSAL driver code upstream for review Philippe Deniel.
>
> In addition, we're working on pNFS support in Ganesha and Ceph, and have a prototype file-based pNFS layout driver for the FSAL that takes advantage of Ceph OSDs for parallel IO.
>
> We're actively extending these efforts and there should be some publication forthcoming.
>
> We have an external git repository here:
>
>    git://github.com/aemerson-linuxbox/ganesha.git
>
> We would really appreciate your feedback--and guidance on how to participate in Ceph, we'd like to.
>
> Matt
>
> ----- "Sage Weil" <sage@newdream.net> wrote:
>
>> Has anyone looked at Genesha?
>>
>>       http://sourceforge.net/apps/trac/nfs-ganesha/
>>
>> In principle it would probably be pretty straightforward to wire that
>> up
>> to libceph and avoid the kernel NFS server for reexporting ceph via
>> NFS...
>>
>> sage
>> --
>> To unsubscribe from this list: send the line "unsubscribe ceph-devel"
>> in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
> --
>
> Matt Benjamin
>
> The Linux Box
> 206 South Fifth Ave. Suite 150
> Ann Arbor, MI  48104
>
> http://linuxbox.com
>
> tel. 734-761-4689
> fax. 734-769-8938
> cel. 734-216-5309
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: userland nfs server

[Brian Chrisman]

On Fri, Jul 29, 2011 at 9:30 AM, Matt W. Benjamin <matt@linuxbox.com> wrote:
> Hi Sage,
>
> We (Linux Box) have built and published a native FSAL for Ceph, which is pretty advanced, and we've submitted the FSAL driver code upstream for review Philippe Deniel.

Looking at the FSAL_CEPH, is this expecting the filesystem to be
mounted via the ceph kernel client?  Have you considered building out
an FSAL using libceph as opposed as to writing directly to OSDs
(assumedly via libRADOS or librbd).

>
> In addition, we're working on pNFS support in Ganesha and Ceph, and have a prototype file-based pNFS layout driver for the FSAL that takes advantage of Ceph OSDs for parallel IO.
>
> We're actively extending these efforts and there should be some publication forthcoming.
>
> We have an external git repository here:
>
>    git://github.com/aemerson-linuxbox/ganesha.git
>
> We would really appreciate your feedback--and guidance on how to participate in Ceph, we'd like to.
>
> Matt
>
> ----- "Sage Weil" <sage@newdream.net> wrote:
>
>> Has anyone looked at Genesha?
>>
>>       http://sourceforge.net/apps/trac/nfs-ganesha/
>>
>> In principle it would probably be pretty straightforward to wire that
>> up
>> to libceph and avoid the kernel NFS server for reexporting ceph via
>> NFS...
>>
>> sage
>> --
>> To unsubscribe from this list: send the line "unsubscribe ceph-devel"
>> in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
> --
>
> Matt Benjamin
>
> The Linux Box
> 206 South Fifth Ave. Suite 150
> Ann Arbor, MI  48104
>
> http://linuxbox.com
>
> tel. 734-761-4689
> fax. 734-769-8938
> cel. 734-216-5309
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: userland nfs server

[Matt W. Benjamin]

Hi Brian,

FSAL_CEPH doesn't involve the kernel client, rather an incidentally modified libceph and libRADOS--it's linked into Ganesha, not using POSIX.  Writing directly to OSDs is part of the pNFS parallel access strategy.

Matt

----- "Brian Chrisman" <brchrisman@gmail.com> wrote:

> On Fri, Jul 29, 2011 at 9:30 AM, Matt W. Benjamin <matt@linuxbox.com>
> wrote:
> > Hi Sage,
> >
> > We (Linux Box) have built and published a native FSAL for Ceph,
> which is pretty advanced, and we've submitted the FSAL driver code
> upstream for review Philippe Deniel.
> 
> Looking at the FSAL_CEPH, is this expecting the filesystem to be
> mounted via the ceph kernel client?  Have you considered building out
> an FSAL using libceph as opposed as to writing directly to OSDs
> (assumedly via libRADOS or librbd).
> 
> >
> > In addition, we're working on pNFS support in Ganesha and Ceph, and
> have a prototype file-based pNFS layout driver for the FSAL that takes
> advantage of Ceph OSDs for parallel IO.
> >
> > We're actively extending these efforts and there should be some
> publication forthcoming.
> >
> > We have an external git repository here:
> >
> >    git://github.com/aemerson-linuxbox/ganesha.git
> >
> > We would really appreciate your feedback--and guidance on how to
> participate in Ceph, we'd like to.
> >
> > Matt
> >
> > ----- "Sage Weil" <sage@newdream.net> wrote:
> >
> >> Has anyone looked at Genesha?
> >>
> >>       http://sourceforge.net/apps/trac/nfs-ganesha/
> >>
> >> In principle it would probably be pretty straightforward to wire
> that
> >> up
> >> to libceph and avoid the kernel NFS server for reexporting ceph
> via
> >> NFS...
> >>
> >> sage
> >> --
> >> To unsubscribe from this list: send the line "unsubscribe
> ceph-devel"
> >> in
> >> the body of a message to majordomo@vger.kernel.org
> >> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >
> > --
> >
> > Matt Benjamin
> >
> > The Linux Box
> > 206 South Fifth Ave. Suite 150
> > Ann Arbor, MI  48104
> >
> > http://linuxbox.com
> >
> > tel. 734-761-4689
> > fax. 734-769-8938
> > cel. 734-216-5309
> > --
> > To unsubscribe from this list: send the line "unsubscribe
> ceph-devel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >

-- 

Matt Benjamin

The Linux Box
206 South Fifth Ave. Suite 150
Ann Arbor, MI  48104

http://linuxbox.com

tel. 734-761-4689
fax. 734-769-8938
cel. 734-216-5309
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: userland nfs server

[Brian Chrisman]

Not sure why this is failing:
AC_CHECK_HEADERS([ceph/libceph.h], [], [AC_MSG_ERROR( missing ceph
include files)])

$ ls -l /usr/include/ceph/libceph.h
-rw-r--r-- 1 root root 7543 Jul 26 10:57 /usr/include/ceph/libceph.h

...
checking for ceph/libceph.h... no
configure: error: missing ceph include files


(I can move this discussion to another ml if preferred.)

On Fri, Jul 29, 2011 at 12:05 PM, Matt W. Benjamin <matt@linuxbox.com> wrote:
> Hi Brian,
>
> FSAL_CEPH doesn't involve the kernel client, rather an incidentally modified libceph and libRADOS--it's linked into Ganesha, not using POSIX.  Writing directly to OSDs is part of the pNFS parallel access strategy.
>
> Matt
>
> ----- "Brian Chrisman" <brchrisman@gmail.com> wrote:
>
>> On Fri, Jul 29, 2011 at 9:30 AM, Matt W. Benjamin <matt@linuxbox.com>
>> wrote:
>> > Hi Sage,
>> >
>> > We (Linux Box) have built and published a native FSAL for Ceph,
>> which is pretty advanced, and we've submitted the FSAL driver code
>> upstream for review Philippe Deniel.
>>
>> Looking at the FSAL_CEPH, is this expecting the filesystem to be
>> mounted via the ceph kernel client?  Have you considered building out
>> an FSAL using libceph as opposed as to writing directly to OSDs
>> (assumedly via libRADOS or librbd).
>>
>> >
>> > In addition, we're working on pNFS support in Ganesha and Ceph, and
>> have a prototype file-based pNFS layout driver for the FSAL that takes
>> advantage of Ceph OSDs for parallel IO.
>> >
>> > We're actively extending these efforts and there should be some
>> publication forthcoming.
>> >
>> > We have an external git repository here:
>> >
>> >    git://github.com/aemerson-linuxbox/ganesha.git
>> >
>> > We would really appreciate your feedback--and guidance on how to
>> participate in Ceph, we'd like to.
>> >
>> > Matt
>> >
>> > ----- "Sage Weil" <sage@newdream.net> wrote:
>> >
>> >> Has anyone looked at Genesha?
>> >>
>> >>       http://sourceforge.net/apps/trac/nfs-ganesha/
>> >>
>> >> In principle it would probably be pretty straightforward to wire
>> that
>> >> up
>> >> to libceph and avoid the kernel NFS server for reexporting ceph
>> via
>> >> NFS...
>> >>
>> >> sage
>> >> --
>> >> To unsubscribe from this list: send the line "unsubscribe
>> ceph-devel"
>> >> in
>> >> the body of a message to majordomo@vger.kernel.org
>> >> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> >
>> > --
>> >
>> > Matt Benjamin
>> >
>> > The Linux Box
>> > 206 South Fifth Ave. Suite 150
>> > Ann Arbor, MI  48104
>> >
>> > http://linuxbox.com
>> >
>> > tel. 734-761-4689
>> > fax. 734-769-8938
>> > cel. 734-216-5309
>> > --
>> > To unsubscribe from this list: send the line "unsubscribe
>> ceph-devel" in
>> > the body of a message to majordomo@vger.kernel.org
>> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> >
>
> --
>
> Matt Benjamin
>
> The Linux Box
> 206 South Fifth Ave. Suite 150
> Ann Arbor, MI  48104
>
> http://linuxbox.com
>
> tel. 734-761-4689
> fax. 734-769-8938
> cel. 734-216-5309
>
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: userland nfs server

[Brian Chrisman]

vinodeno_t is not accessible via libceph (was it previously?) though
it's still used and it's in mdstypes.h, though it doesn't look like
it'll include in a straight C compile.  Is there a missing include?

[bchrisman (aemerson): src] $ find . -name '*.c' -o -name '*.h' |
xargs grep inodeno
./include/FSAL/FSAL_CEPH/fsal_types.h:   vinodeno_t vi;
./include/FSAL/FSAL_CEPH/fsal_types.h:  vinodeno_t vi;
./include/FSAL/FSAL_CEPH/fsal_types.h:  vinodeno_t vi;
[bchrisman (aemerson): src] $

../../include/FSAL/FSAL_CEPH/fsal_types.h:73: error: expected
specifier-qualifier-list before ‘vinodeno_t’

from mdstypes:

struct vinodeno_t {
  inodeno_t ino;
  snapid_t snapid;
  vinodeno_t() {}
  vinodeno_t(inodeno_t i, snapid_t s) : ino(i), snapid(s) {}

  void encode(bufferlist& bl) const {
    ::encode(ino, bl);
    ::encode(snapid, bl);
  }
  void decode(bufferlist::iterator& p) {
    ::decode(ino, p);
    ::decode(snapid, p);
  }
};
WRITE_CLASS_ENCODER(vinodeno_t)

On Fri, Jul 29, 2011 at 12:05 PM, Matt W. Benjamin <matt@linuxbox.com> wrote:
> Hi Brian,
>
> FSAL_CEPH doesn't involve the kernel client, rather an incidentally modified libceph and libRADOS--it's linked into Ganesha, not using POSIX.  Writing directly to OSDs is part of the pNFS parallel access strategy.
>
> Matt
>
> ----- "Brian Chrisman" <brchrisman@gmail.com> wrote:
>
>> On Fri, Jul 29, 2011 at 9:30 AM, Matt W. Benjamin <matt@linuxbox.com>
>> wrote:
>> > Hi Sage,
>> >
>> > We (Linux Box) have built and published a native FSAL for Ceph,
>> which is pretty advanced, and we've submitted the FSAL driver code
>> upstream for review Philippe Deniel.
>>
>> Looking at the FSAL_CEPH, is this expecting the filesystem to be
>> mounted via the ceph kernel client?  Have you considered building out
>> an FSAL using libceph as opposed as to writing directly to OSDs
>> (assumedly via libRADOS or librbd).
>>
>> >
>> > In addition, we're working on pNFS support in Ganesha and Ceph, and
>> have a prototype file-based pNFS layout driver for the FSAL that takes
>> advantage of Ceph OSDs for parallel IO.
>> >
>> > We're actively extending these efforts and there should be some
>> publication forthcoming.
>> >
>> > We have an external git repository here:
>> >
>> >    git://github.com/aemerson-linuxbox/ganesha.git
>> >
>> > We would really appreciate your feedback--and guidance on how to
>> participate in Ceph, we'd like to.
>> >
>> > Matt
>> >
>> > ----- "Sage Weil" <sage@newdream.net> wrote:
>> >
>> >> Has anyone looked at Genesha?
>> >>
>> >>       http://sourceforge.net/apps/trac/nfs-ganesha/
>> >>
>> >> In principle it would probably be pretty straightforward to wire
>> that
>> >> up
>> >> to libceph and avoid the kernel NFS server for reexporting ceph
>> via
>> >> NFS...
>> >>
>> >> sage
>> >> --
>> >> To unsubscribe from this list: send the line "unsubscribe
>> ceph-devel"
>> >> in
>> >> the body of a message to majordomo@vger.kernel.org
>> >> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> >
>> > --
>> >
>> > Matt Benjamin
>> >
>> > The Linux Box
>> > 206 South Fifth Ave. Suite 150
>> > Ann Arbor, MI  48104
>> >
>> > http://linuxbox.com
>> >
>> > tel. 734-761-4689
>> > fax. 734-769-8938
>> > cel. 734-216-5309
>> > --
>> > To unsubscribe from this list: send the line "unsubscribe
>> ceph-devel" in
>> > the body of a message to majordomo@vger.kernel.org
>> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> >
>
> --
>
> Matt Benjamin
>
> The Linux Box
> 206 South Fifth Ave. Suite 150
> Ann Arbor, MI  48104
>
> http://linuxbox.com
>
> tel. 734-761-4689
> fax. 734-769-8938
> cel. 734-216-5309
>
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Fwd: userland nfs server

[Adam C. Emerson]

> vinodeno_t is not accessible via libceph (was it previously?) though
> it's still used and it's in mdstypes.h, though it doesn't look like
> it'll include in a straight C compile.  Is there a missing include?

Sorry about that.  We made a few changes to the ceph client library,
which can be found at git://github.com/aemerson-linuxbox/ceph.git

I recommend repulling NFS-ganesha and using the beforelock branch, for
the moment.

Re: userland nfs server

[Sage Weil]

Hi Matt,

On Fri, 29 Jul 2011, Matt W. Benjamin wrote:
> Hi Sage,
> 
> We (Linux Box) have built and published a native FSAL for Ceph, which is 
> pretty advanced, and we've submitted the FSAL driver code upstream for 
> review Philippe Deniel.

Not sure if I said this before, but this is awesome!

> In addition, we're working on pNFS support in Ganesha and Ceph, and have 
> a prototype file-based pNFS layout driver for the FSAL that takes 
> advantage of Ceph OSDs for parallel IO.
> 
> We're actively extending these efforts and there should be some 
> publication forthcoming.
> 
> We have an external git repository here:
> 
>     git://github.com/aemerson-linuxbox/ganesha.git

I just took a look at your ceph.git and it looks like there are several 
changes you need to make to libceph.  It also looks like it's based on 
v0.27 or thereabouts, which is pretty old now.  There seems to be a mix of 
pretty simple/straighforward things that are easily merged upstream, and 
some larger API additions that I have some questions about.

Are you guys actively working on this?  Can we work together to get these 
libceph API changes merged into the mainline?  I would love to see this 
run against the packaged libceph (and get the ceph FSAL upstream into 
ganesha).

Thanks!
sage



> 
> We would really appreciate your feedback--and guidance on how to 
> participate in Ceph, we'd like to.
> 
> Matt
> 
> ----- "Sage Weil" <sage@newdream.net> wrote:
> 
> > Has anyone looked at Genesha?
> > 
> > 	http://sourceforge.net/apps/trac/nfs-ganesha/
> > 
> > In principle it would probably be pretty straightforward to wire that
> > up 
> > to libceph and avoid the kernel NFS server for reexporting ceph via
> > NFS...
> > 
> > sage
> > --
> > To unsubscribe from this list: send the line "unsubscribe ceph-devel"
> > in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> -- 
> 
> Matt Benjamin
> 
> The Linux Box
> 206 South Fifth Ave. Suite 150
> Ann Arbor, MI  48104
> 
> http://linuxbox.com
> 
> tel. 734-761-4689
> fax. 734-769-8938
> cel. 734-216-5309
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> 

Re: userland nfs server

[Matt W. Benjamin]

Hi Sage,

Yes, we're actively working on this.  We'd very much like assistance adapting the libceph changes for inclusion.  We can start by updating for recent Ceph.

FSAL_CEPH will definitely merge in Ganesha mainline in the next month or two.

Matt

----- "Sage Weil" <sage@newdream.net> wrote:

> Hi Matt,
> 
> On Fri, 29 Jul 2011, Matt W. Benjamin wrote:
> > Hi Sage,
> > 
> > We (Linux Box) have built and published a native FSAL for Ceph,
> which is 
> > pretty advanced, and we've submitted the FSAL driver code upstream
> for 
> > review Philippe Deniel.
> 
> Not sure if I said this before, but this is awesome!
> 
> > In addition, we're working on pNFS support in Ganesha and Ceph, and
> have 
> > a prototype file-based pNFS layout driver for the FSAL that takes 
> > advantage of Ceph OSDs for parallel IO.
> > 
> > We're actively extending these efforts and there should be some 
> > publication forthcoming.
> > 
> > We have an external git repository here:
> > 
> >     git://github.com/aemerson-linuxbox/ganesha.git
> 
> I just took a look at your ceph.git and it looks like there are
> several 
> changes you need to make to libceph.  It also looks like it's based on
> 
> v0.27 or thereabouts, which is pretty old now.  There seems to be a
> mix of 
> pretty simple/straighforward things that are easily merged upstream,
> and 
> some larger API additions that I have some questions about.
> 
> Are you guys actively working on this?  Can we work together to get
> these 
> libceph API changes merged into the mainline?  I would love to see
> this 
> run against the packaged libceph (and get the ceph FSAL upstream into
> 
> ganesha).
> 
> Thanks!
> sage
> 
> 
> 
> > 
> > We would really appreciate your feedback--and guidance on how to 
> > participate in Ceph, we'd like to.
> > 
> > Matt
> > 
> > ----- "Sage Weil" <sage@newdream.net> wrote:
> > 
> > > Has anyone looked at Genesha?
> > > 
> > > 	http://sourceforge.net/apps/trac/nfs-ganesha/
> > > 
> > > In principle it would probably be pretty straightforward to wire
> that
> > > up 
> > > to libceph and avoid the kernel NFS server for reexporting ceph
> via
> > > NFS...
> > > 
> > > sage
> > > --
> > > To unsubscribe from this list: send the line "unsubscribe
> ceph-devel"
> > > in
> > > the body of a message to majordomo@vger.kernel.org
> > > More majordomo info at 
> http://vger.kernel.org/majordomo-info.html
> > 
> > -- 
> > 
> > Matt Benjamin
> > 
> > The Linux Box
> > 206 South Fifth Ave. Suite 150
> > Ann Arbor, MI  48104
> > 
> > http://linuxbox.com
> > 
> > tel. 734-761-4689
> > fax. 734-769-8938
> > cel. 734-216-5309
> > --
> > To unsubscribe from this list: send the line "unsubscribe
> ceph-devel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > 
> > 
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel"
> in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

-- 

Matt Benjamin

The Linux Box
206 South Fifth Ave. Suite 150
Ann Arbor, MI  48104

http://linuxbox.com

tel. 734-761-4689
fax. 734-769-8938
cel. 734-216-5309

Re: userland nfs server

[Koffi Nogbe]

>Matt W. Benjamin <matt <at> linuxbox.com> writes:

> 
> Hi Sage,
> 
> Yes, we're actively working on this.  We'd very much like assistance
adapting the libceph changes for
> inclusion.  We can start by updating for recent Ceph.
> 
> FSAL_CEPH will definitely merge in Ganesha mainline in the next month or two.
> 
> Matt
> 
> ----- "Sage Weil" <sage <at> newdream.net> wrote:
> 



>Matt W. Benjamin <matt <at> linuxbox.com> writes:

I m trying to build the CEPH FSAL but apparently I m failing. I was able to
pass the first issue about vinode struct that was resolved by using the ceph
source from linuxbox but I m stopped by this one that I cannot figure out:

/bin/sh ../../libtool --tag=CC   --mode=compile gcc -DHAVE_CONFIG_H -I.
-I../../include    -D_FILE_OFFSET_BITS=64  -I/usr/include
-D_FILE_OFFSET_BITS=64 -I/root/nfs-ganesha-1.5.1/libtirpc/tirpc/ -g -O2
-D_REENTRANT -Wall -Wimplicit -Wformat -Wmissing-braces -Wno-pointer-sign
-I../../RPCAL/gssd -MT fsal_dirs.lo -MD -MP -MF .deps/fsal_dirs.Tpo -c -o
fsal_dirs.lo fsal_dirs.c
libtool: compile:  gcc -DHAVE_CONFIG_H -I. -I../../include
-D_FILE_OFFSET_BITS=64 -I/usr/include -D_FILE_OFFSET_BITS=64
-I/root/nfs-ganesha-1.5.1/libtirpc/tirpc/ -g -O2 -D_REENTRANT -Wall
-Wimplicit -Wformat -Wmissing-braces -Wno-pointer-sign -I../../RPCAL/gssd
-MT fsal_dirs.lo -MD -MP -MF .deps/fsal_dirs.Tpo -c fsal_dirs.c  -fPIC -DPIC
-o .libs/fsal_dirs.o
fsal_dirs.c: In function âCEPHFSAL_readdirâ:
fsal_dirs.c:166: error: cannot convert to a pointer type
fsal_dirs.c:166: error: âcephfsal_cookie_tâ has no member named âcookieâ
fsal_dirs.c:167: error: cannot convert to a pointer type
fsal_dirs.c:167: error: âcephfsal_cookie_tâ has no member named âcookieâ
make[2]: *** [fsal_dirs.lo] Error 1
make[2]: Leaving directory `/root/nfs-ganesha-1.5.1/FSAL/FSAL_CEPH'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/root/nfs-ganesha-1.5.1/FSAL'
make: *** [all-recursive] Error 1




--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: userland nfs server

[Matt W. Benjamin]

Hi Koffi,

It looks like you're building the Ganesha "stable" branch.  You want to be using the
pre-2.0 dev branch--that's the "next" branch from https://github.com/nfs-ganesha/nfs-ganesha.git .

There's some limitations with pNFS in this branch, but you can experiment with ordinary
NFSv3, NFSv4, and 4.1 exports.

Thanks,

Matt

----- "Koffi Nogbe" <knogbe@gmail.com> wrote:

> >Matt W. Benjamin <matt <at> linuxbox.com> writes:
> 
> > 
> > Hi Sage,
> > 
> > Yes, we're actively working on this.  We'd very much like
> assistance
> adapting the libceph changes for
> > inclusion.  We can start by updating for recent Ceph.
> > 
> > FSAL_CEPH will definitely merge in Ganesha mainline in the next
> month or two.
> > 
> > Matt
> > 
> > ----- "Sage Weil" <sage <at> newdream.net> wrote:
> > 
> 
> 
> 
> >Matt W. Benjamin <matt <at> linuxbox.com> writes:
> 
> I m trying to build the CEPH FSAL but apparently I m failing. I was
> able to
> pass the first issue about vinode struct that was resolved by using
> the ceph
> source from linuxbox but I m stopped by this one that I cannot figure
> out:
> 
> /bin/sh ../../libtool --tag=CC   --mode=compile gcc -DHAVE_CONFIG_H
> -I.
> -I../../include    -D_FILE_OFFSET_BITS=64  -I/usr/include
> -D_FILE_OFFSET_BITS=64 -I/root/nfs-ganesha-1.5.1/libtirpc/tirpc/ -g
> -O2
> -D_REENTRANT -Wall -Wimplicit -Wformat -Wmissing-braces
> -Wno-pointer-sign
> -I../../RPCAL/gssd -MT fsal_dirs.lo -MD -MP -MF .deps/fsal_dirs.Tpo -c
> -o
> fsal_dirs.lo fsal_dirs.c
> libtool: compile:  gcc -DHAVE_CONFIG_H -I. -I../../include
> -D_FILE_OFFSET_BITS=64 -I/usr/include -D_FILE_OFFSET_BITS=64
> -I/root/nfs-ganesha-1.5.1/libtirpc/tirpc/ -g -O2 -D_REENTRANT -Wall
> -Wimplicit -Wformat -Wmissing-braces -Wno-pointer-sign
> -I../../RPCAL/gssd
> -MT fsal_dirs.lo -MD -MP -MF .deps/fsal_dirs.Tpo -c fsal_dirs.c  -fPIC
> -DPIC
> -o .libs/fsal_dirs.o
> fsal_dirs.c: In function âCEPHFSAL_readdirâ:
> fsal_dirs.c:166: error: cannot convert to a pointer type
> fsal_dirs.c:166: error: âcephfsal_cookie_tâ has no member named
> âcookieâ
> fsal_dirs.c:167: error: cannot convert to a pointer type
> fsal_dirs.c:167: error: âcephfsal_cookie_tâ has no member named
> âcookieâ
> make[2]: *** [fsal_dirs.lo] Error 1
> make[2]: Leaving directory `/root/nfs-ganesha-1.5.1/FSAL/FSAL_CEPH'
> make[1]: *** [all-recursive] Error 1
> make[1]: Leaving directory `/root/nfs-ganesha-1.5.1/FSAL'
> make: *** [all-recursive] Error 1
> 
> 
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel"
> in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

-- 
Matt Benjamin
The Linux Box
206 South Fifth Ave. Suite 150
Ann Arbor, MI  48104

http://linuxbox.com

tel.  734-761-4689 
fax.  734-769-8938 
cel.  734-216-5309 
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

userland nfs server

[Sage Weil]

Has anyone looked at Genesha?

	http://sourceforge.net/apps/trac/nfs-ganesha/

In principle it would probably be pretty straightforward to wire that up 
to libceph and avoid the kernel NFS server for reexporting ceph via NFS...

sage