Hello Andy
I was reading the oss/ofs documentation last night and also thought
that if I would write my own mmssgwcmd I could prevent xrootd from using
the MSS. As you suggested (prefer) I will write a little script that
implements the mmssgwcmd for the local file system.
There still might be some issues because we use a mixed setting on the
bbrprod machine: the /prod area is using only the cache system but no
HPSS, /store is using the cache system, does migration and might
want to use a mmssgwcmd that access hpss. I have to think a little
bit about this and what the implications are.
Cheers,
Wilko
On Tue, 23 Nov 2004, Andrew Hanushevsky wrote:
> Hi Wilko,
>
> OK, it appears that you can't have your cake and eat it too. If you want
> the special files to be created in the file system but don't want to use
> the MSS for anything then you will need to write your own mmssgwcmd that
> does the operations locally instead of remotely. This completely
> documented in the ofs/oss manual under mssgwcmd. While it would be
> possible to add an option to indicate that only special files shuld be
> created and, otherwise, no MSS is being used; it would be faster to just
> write a small perl script that implements the various commands that would
> be directed to the MSS.
>
> It is apparent that dirlist should be done only locally (that that will
> not give you the complete list of files but only those on disk). The rm,
> mv, and create commands should probably do nothing more that return a zero
> status code. I assume you also want these to occur locally. The statx
> command, well, you have to decide what to do there. This gets called if
> the file/dir does not exist on the local disk.
>
> Let me know which way you want to proceed. If you would like to write this
> (prefered) or you want me to write the script.
>
> Andy
>
> On Mon, 22 Nov 2004, Wilko Kroeger wrote:
>
> >
> > Hello Remi
> >
> > Your problem is due to the configuration of the xrootd servers.
> > Trying to get an directory listing (/kanga/prod/s5/s4/s3/s2/s1 on
> > bbrprod05), I see in the xrootd log file:
> >
> > 041122 23:34:55 059 XrootdXeq: User logged in as wilko.6750:18@noric06
> > pudc: Error opening directory /kanga/prod/s5/s4/s3/s2/s1/; No such file or directory.
> > 041122 23:34:55 059 oss_XrdOssMSS_Xeq: Unable to execute dlist; no such file or directory
> > 041122 23:34:55 059 ofs_opendir: wilko.6750:18@noric06 Unable to open
> > directory /prod/s5/s4/s3/s2/s1/; No such file or directory
> >
> > It looks to me that xrootd tries to check the file in HPSS and not on
> > disk. I tried to get a dir listing from '/store/' and I get a
> > directory listing from hpss.
> >
> > Andy: How to configure xrootd so that dir listings are done from a disk
> > and not HPSS?
> > Another question comes to my mind: What happens if I remove a file or
> > directory? Does it get removed from HPSS ?
> >
> >
> > Cheers,
> > Wilko
> >
> >
> >
> > On Mon, 22 Nov 2004, Remi Mommsen wrote:
> >
> > > Hi,
> > >
> > > I'm using today's cvs HEAD for XrdClientAdmin.
> > >
> > > I try to get a directory listing using the perl interface like:
> > >
> > > XrdClientAdmin::XrdInitialize("root://bbrprod05:1094/dummy", 0);
> > > XrdClientAdmin::XrdDirList('/prod/sub5/');
> > >
> > > but I get:
> > > 041122 22:25:03 001 Xrd: DirList Server
> > > [bbrprod05.slac.stanford.edu:1094] did not return OK message for last
> > > request.
> > > 041122 22:25:03 001 Xrd: SendGenCommand Server declared error
> > > 3011:Unable to open directory /prod/sub5/; No such file or directory
> > >
> > > while the directory is there:
> > > [bbrprod05] ~ > ls -l /kanga/prod/sub5/
> > > total 2
> > > drwxr-sr-x 3 bbdatsrv bfactory 512 Nov 22 22:17 sub6
> > >
> > >
> > >
> > > Removing the directory fails, too:
> > > XrdClientAdmin::XrdRmdir("/prod/sub5/sub6/sub7/");
> > >
> > > 041122 22:26:06 001 Xrd: Rmdir Server
> > > [bbrprod05.slac.stanford.edu:1094] did not return OK message for last
> > > request.
> > > 041122 22:26:06 001 Xrd: SendGenCommand Server declared error
> > > 3010:Unable to remove /prod/sub5/sub6/sub7/; Not owner
> > >
> > >
> > > Is this a server configuration problem (it occurs on all bbrprod0X
> > > machines at SLAC), or is there a bug in the code?
> > >
> > > Cheers,
> > > Remi
> > >
> > >
> > > ---------------------------------------------------------------------
> > > Truth decays into beauty, while beauty soon becomes merely charm.
> > > Charm ends up as strangeness, and even that doesn't last, but up and
> > > down are forever. (Anonymous)
> > >
> > > *********************************************************************
> > > Remigius K. Mommsen e-mail: [log in to unmask]
> > > University of California, Irvine URL: http://cern.ch/mommsen
> > > c/o SLAC voice: ++1 (650) 926-3595
> > > 2575 Sand Hill Road #35 fax: ++1 (650) 926-3882
> > > Menlo Park, CA 94025, US home: ++1 (650) 233-9041
> > > *********************************************************************
> > >
> > >
> >
>
