Print

Print


On Wed, Sep 21, 2005 at 12:26:36PM +0100 or thereabouts, William Hay wrote:
> > Hi,
> > 
> >   + UCL 
> >     Submission error but I can't reproduce at the moment so maybe fixed
> >     this morning. Looking at the error in the SFT this would be easy
> >     to spot if it were still there.
> We've been looking into this at UCL
> 
> This problem seems to occur with the SFTs but not with jobs we submit ourselves
> via the RAL RB.
> When we submit jobs via the RAL RB they work fine.
> However when the RBs at CERN try to connect we get this filling up our
> gatekeeper logs:
> 
> >Notice: 6: Got connection 137.138.154.178 at Wed Sep 21 11:14:01 2005
> >
> >Notice: 6: Got connection 137.138.154.177 at Wed Sep 21 11:14:01 2005
> >
> >Failed reading length 0
> >GSS authentication failure
> >    globus_gss_assist token :3: read failure: Connection closed
> >    Failure: GSS failed Major:01090000 Minor:00000000 Token:00000003
> >
> >    Failure: GSS failed Major:01090000 Minor:00000000 Token:00000003
> >
> >    Failed reading length 0
> >    GSS authentication failure
> >        globus_gss_assist token :3: read failure: Connection closed
> >	Failure: GSS failed Major:01090000 Minor:00000000 Token:00000003
> >
> >	Failure: GSS failed Major:01090000 Minor:00000000 Token:00000003
> >
> >	Notice: 6: Got connection 137.138.154.177 at Wed Sep 21 11:14:25 2005
> >
> >	Notice: 6: Got connection 137.138.154.177 at Wed Sep 21 11:14:30 2005
> >
> >	Failed reading length 0
> >	GSS authentication failure
> >	    globus_gss_assist token :3: read failure: Connection closed
> >	    Failure: GSS failed Major:01090000 Minor:00000000 Token:00000003
> >
> >	    Failure: GSS failed Major:01090000 Minor:00000000 Token:00000003
> The IP addresses in our logs are those of gdrb01.cern.ch and gdrb02.cern.ch
> 
> Our gridmapfile contains an entry for the user submitting the SFTs so presumably
> the should be authenticated.  Googling for the GSS Failed line turned up a few things
> but nothing that appears to match closely although they suggest there is at least a
> 50/50 chance the problem is at the submitters end.
> 
> Any suggestions?
> 
> William

Something odd going on here 

lcgui01.gridpp.rl.ac.uk>  edg-gridftp-ls gsiftp://ce-a.ccc.ucl.ac.uk/
/etc/
...

lxplus.cern.ch> edg-gridftp-ls gsiftp://ce-a.ccc.ucl.ac.uk/
Error 5: couldn't register read buffer

which is a new one o me.

  Steve



-- 
Steve Traylen
[log in to unmask]
http://www.gridpp.ac.uk/