[Swift-user] Success with fork, but exception in getFile with condor

Mihael Hategan hategan at mcs.anl.gov
Tue Sep 18 16:08:18 CDT 2007


On Mon, 2007-09-17 at 15:09 -0500, Anand Padmanabhan wrote:
> >>
> >>>  Also I know some condor systems, job executables get dumped a 
> >>> temporary directory on a worker node's local file system. Would this 
> >>> have any effect on Swift?
> >>
> >> As long as Condor/the job manager honor the directory rls setting, this
> >> shouldn't make any difference.
> > This is something we need to make sure this is the case. I know we had a 
> > earlier problem at FNAL_FERMIGRID on which the initial dir globus 
> > parameter was not respected. You can find details at 
> > https://twiki.grid.iu.edu/twiki/bin/view/Troubleshooting/NewUserRunningJobsFailureFNAL 
> I checked with FNAL with 2 of the sites on which Jing was having 
> problems with. As I suspected the siteadmin confirmed that the two 
> gatekeepers in question were running NFS-lite and do not respect the RSL 
> initial directory variable. The lack of support of initialdir on some 
> OSG sites is known issue and it does break compatibility.
> 
> Also, is there a way (some parameter we specify), so that swift not to 
> set the initialdir parameter. This way the job can finish in what ever 
> directory it gets dumped to by the batch system and then possibly copy 
> files over to expected location in the $OSG_DATA directory.

No. There is no such parameter yet.

Is not implementing random bits of an otherwise standard interface an
acceptable thing on OSG? Are there any other "surprises" we should be
aware of?

Mihael

> 
> Thanks,
> Anand
> 




More information about the Swift-user mailing list