2009/12/16 Anton Starikov <[email protected]>: > > On Dec 16, 2009, at 7:08 PM, Jeremy Allison wrote: > >> On Wed, Dec 16, 2009 at 07:00:09PM +0100, Anton Starikov wrote: >>> And although it creates directory, it doesn't copy contents, because it >>> stops process of copying directory after this error. If I repeat filesync, >>> the contents of directory will be copid (cause directory is already here). >>> >>> So, it looks exactly the same. >>> If so, then problem in chflags(). >>> I expect that samba on linux is compiled without support for chflags, >>> obviously. >>> >>> I presume that settings "unix extensions = no" would probably fix this, but >>> it has a drawback, because then you loose native unix things like symlinks >>> etc. >>> >>> Which is, at least in our case is not possible, cause shares accessed by >>> both, mac and linux clients over NFS (the same clients on different hosts) >>> and symlinks are heavily used. >>> >>> I think, OSX client, when it sees that server supports "unix extensions", >>> expects that on other side is OSX server with samba which supports chflags. >>> >>> So, if we don't discuss rewrite of OSX cifs FS, then only solution is to >>> "emulate" chflags support on samba side (or convert flags to XFS/ETX3 attrs >>> somehow) >> >> Hmmm. Looks like a client bug then, in that they don't cope with an >> error on chflags set. What error is the Samba server returning here ? > > Of course it is client error. But it is much easy to add "dirty hack" to > samba on server that fooling around bunch of clients. > Does Apple opensource their implementation of smbfs?
http://www.opensource.apple.com/source/smb/smb-348.7/ -- James Peach | [email protected] -- To unsubscribe from this list go to the following URL and read the instructions: https://lists.samba.org/mailman/options/samba
