[GE users] Bizarre Mixing Of Users Output/Error Files

Jon Forrest jlforrest at berkeley.edu
Fri Nov 14 19:58:22 GMT 2008

Geoff Galitz wrote:
> Hi Jon.
> FWIW, I've seen this not under SGE and Rocks, specifically, but 
> I have seen similar issues when the filesystem is damaged.  Perhaps
> a full fsck check is in order or a RAID verify?

This is on an XFS file system. I don't think I could
do fsck on a live filesystem. I know the RAID unit
is fine. Unfortunately, Adam Willard removed the files
before I could see them (Adam just got his PhD last

> I suppose it is also possible that if the permissions are wide open 
> that someone with a poorly written submission script can have 
> this affect.  It is fairly common for folks to run in-house code 
> from someone else's private bin directory.

Not in this case. Each user's home directory has 700 permission
so they don't have access to each other's directory.

This is on a new 20 node dual-quad core cluster that
I just set up for Chandler. It works great! I also
converted Jhih-Wei's, Birgitta's, and Kathy's cluster
to Rocks. Also, Kathy just got a proposal for a new
cluster ($400K) funded so we're either going to replace
bedrock or add a new cluster somewhere.

Other than the crazy financial situation here things
are going pretty well.



To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].

More information about the gridengine-users mailing list