This is the mail archive of the
cygwin-apps@cygwin.com
mailing list for the Cygwin project.
RE: NFS server (final?)
- From: Pavel Tsekov <ptsekov at gmx dot net>
- To: cygwin-apps at cygwin dot com
- Date: Fri, 14 Feb 2003 11:19:29 +0100 (CET)
- Subject: RE: NFS server (final?)
Hello,
I've just tried the new binaries and the nfs daemon no longer fills the
Application Log with error messages :)
The other problem is still there, but this is expected. If anyone can
offer any hints on how to debug this problem (I'm not familiar
with nfsd at all) I'd be happy to help to trace it down. Btw it is 100%
reproducible in my setup and with the suggested file sizes.
Sam, some notes about the Cygwin specific readme file:
Maybe you have to add some lines to it as you have suggested:
"So, is the answer to remove the seteuid(ROOT_UID) call, and document
the requirement that the server be run under an account with the "Create
a process level token" right?"
Also I noticed the following minor inconsistencies in the readme file:
Runtime requirements:
cygwin-1.3.17 or newer
Build requirements:
cygwin-1.3.17 or newer
sunrpc
Maybe cygwin-1.3.20 would be better. If you're actually using 1.3.17 then
this is OK.
Also this line is incorrect:
Files included in the binary distro
/usr/bin/nfs-server-config
/usr/doc/Cygwin/nfs-server-2.3-1.README
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Having said that, I think it's time to decide if we want the nfs
server to become a part of the Cygwin distro now. If we put aside the
problem with the directory removal maybe it is good to release this
version as is. Then when the "bug" is fixed a version -2 will be released.
I don't really know, how big is the demand of a nfs server and how
the inclusion of one will impact the popularity of Cygwin, but I remember
some comments on the main list, that some people are using Interix only
for its nfs server functionality.
Any comments ?
On Thu, 13 Feb 2003, Robb, Sam wrote:
> > Thanks you very much! I'll make the changes, rebuild and
> > post new binaries & source today.
>
> Done. Current version no longer tries to seteuid(0) before changing
> users. I haven't had a chance to look at the other problem Pavel has
> reported w/regard to file/directory deletes, though.