A user's job seg-faulted when running in our system and produced the attached stack trace. 
  The information in the stack trace indicates that the problems seemed to occur when the 
xrootd client was trying to allocate additional memory (Thread 1).  The code should have 
been using the client libraries associated with the ROOT version packaged with ATLAS 
release.  The root version is labelled as "Version  5.26/00e   13 October 2010"

I am curious if my diagnosis is correct, and if anyone has seen something similar before?

I reran the job on separate machine and it seemed to run fine.