What should I do if my Jaguar job fails with the following error message (or something very similar):

"Record 24 of IO unit 81 (ionbyn) is empty"

Several hundred megabytes of free disk space are needed to run most Jaguar jobs, and it is possible that for a large job, more than a gigabyte will be required. If you receive a message similar to the one above, you do not have enough scratch disk space for Jaguar to complete the attempted calculation. It could be that adequate space was not allocated, or that the scratch disk was being used simultaneously by other processes, perhaps by Jaguar jobs of other users.

To determine if the allocated space is adequate, on Linux you can go to your scratch directory and type

df -k

at the prompt. The current usage for that directory will be displayed. On Windows, go to My Computer, right-click on the drive used for scratch files, and choose Properties. The usage is listed in the General tab.

The scratch directory that Jaguar uses is determined by the tmpdir setting for the host in the schrodinger.hosts file. For jobs run locally, the entry for localhost is used. If no tmpdir is specified, then Jaguar uses the default location, $HOME/.schrodinger/tmp on Linux, and %LOCALAPPDATA%\Schrodinger\tmp on Windows.

Back to Search Results