[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] NFS errors with log file

On Thu, 5 Jun 2008, Brent Strong wrote:

I just updated some config settings to reflect our move to NFS-based
home directories and when testing an old job, I'm getting this error:

WARNING: Can't determine whether log file /home/testhome/Hello.log is
on NFS

Is there something else I need to update in the condor_config file or
is it an issue with the job itself?

Well, ideally you'd change your submit files to move your log files off of
NFS!  What's going on here is that you should get a warning if you submit
a job whose log file is on NFS, but the code that tries to determine
whether the log file is on NFS is having problems doing so.

The reason for all of this is that file locking generally doesn't work reliably on NFS, which can cause corruption of the log files. For stand-alone jobs, having a log file on NFS generates a warning, but for a DAGMan node job, it generates a fatal error (because the log files are crucial to the proper execution of the DAG). (You can change this behavior with the configuration macros mentioned below.)

You should look at the information for the LOG_ON_NFS_IS_ERROR and DAGMAN_LOG_ON_NFS_IS_ERROR configuration macros in the configuration section of the manual (http://www.cs.wisc.edu/condor/manual/v7.0/3_3Configuration.html) for 7.0.

Kent Wenger
Condor Team