[Lustre-discuss] Questions on patch-less client verses patched kernel Lustre implementations.

[Eric Adint]

I am trying to diagnose some instability problems that we are having with our lustre installation here is a breakdown of our configuration 
OFED 1.3 
Rocks 5.3 
Lustre 1.8.2 
Kernel 2.6.18-164.11.1.el5

On the Lustre servers we are running the patched full version of Lustre 
on the clients we are running the patchless client. 
One of the problems we are seeing is a high degree of instability in the file system, disconnects and file access problems 
and a verry high degree of Transport endpoint shutdowns on both the IP and IB side of the house. 
Has anyone else seen these kind of problems and is it better run the patched kernel full version on the clients, or what is the difference between the patch-less and patched client as far as stability goes. 
Finally, will Lustre 1.8.3 help with many of these problems. 

Eric Adint
ehadint at nps.edu
HPC specialist  Research Computing  
Naval Postgraduate School
833 Dyer Road Bldg 232 Room 139a
Monterey Ca 93943
831-402-5996