[OpenAFS-devel] Re: on sockets and devices in AFS

Spencer E. Olson olsonse@umich.edu
Thu, 16 Sep 2010 10:59:06 -0600


=46rom the users' point of view, practically speaking, another advantage fo=
r not=20
writing these back is that a user could start up processes that create thes=
e=20
sockets/pipes from multiple locations without disruption. =20

If you have sockets/named pipes written back to the server, can they really=
 be=20
used from multiple clients?  The idea of "distributed sockets/named pipes" =
as=20
Tom Keiser pointed out would require implementation in the file server,  i.=
e.=20
socket/pipe data to be sent accross the file server (wouldn't it?).  It als=
o=20
seems that this kind of solution would require some new type of knowledge o=
n=20
the part of the software that reads/writes to/from the sockets/pipes.  This=
=20
seems a bit more of a stretch in terms of purpose/goals of AFS.

On Thursday 16 September 2010 10:32, Derrick Brashear wrote:
> The point of not writing back is that it can be done today without protoc=
ol
> change. No issue of what unsuspecting clients see, because what they see =
is
> ... nothing.
>
> On Sep 16, 2010, at 3:53 PM, Andrew Deason <adeason@sinenomine.net> wrote:
> > On Thu, 16 Sep 2010 07:53:59 -0400
> >
> > chas williams - CONTRACTOR <chas@cmf.nrl.navy.mil> wrote:
> >> On Thu, 16 Sep 2010 09:58:22 +0200
> >>
> >> Derrick Brashear <shadow@dementia.org> wrote:
> >>> 2) if created locally they would cover a later file of the same name
> >>> if one is created
> >>
> >> how about a new 'file type' that is 'special' just like mounts.
> >> however, they would only have a local (to the client host)
> >> meaning.  old clients would see 'files' (with a strange prefix i
> >> guess) and new clients would be able to handle these things as special
> >> devices (named pipes and the like).
> >
> > Old clients could write stuff to that file, though, which is also thus
> > hidden from new clients. Unless you add knowledge about the special
> > prefix or whatever to the fileserver, prohibiting such.
> >
> > We can't create an actual new file type? (as in, an
> > AFSFetchStatus.FileType type)
> >
> > --
> > Andrew Deason
> > adeason@sinenomine.net
> >
> > _______________________________________________
> > OpenAFS-devel mailing list
> > OpenAFS-devel@openafs.org
> > https://lists.openafs.org/mailman/listinfo/openafs-devel
>
> _______________________________________________
> OpenAFS-devel mailing list
> OpenAFS-devel@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-devel