All – I want to give you a update on the filesystem issue we experienced with the scratch filesystem on Discovery last night.  While it has been addressed for now, it is useful to understand what happened and what you can do to help us avoid this issue in the future.

 

Filesystems like /gss_gpfs_scratch have limits on available space (which we have discussed in the recent past) as well as on the number of file/directory locators or pointers (called inodes) that can be stored.  The inodes table essentially keeps track of every file and directory on the file system.  Every file and directory on a disk requires one inode entry as well as allocated physical storage space on which to save the data.

 

Last night, even though there was physical storage space available, the /gss_gpfs_scratch filesystem ran into the configured limit on the inode table.  This caused the filesystem to be unable to process requests for new file/directory creation, and you may have experienced errors in your jobs in this respect.   The ITS Research Computing Support team jumped on this issue as soon as we saw what was going on, diagnosed and identified the problem, and took various measures to alleviate the problem _immediately_.  The inode table limit has been increased to the maximum number available in the filesystem.

 

At the moment, the /gss_gpfs_scratch partition is 68% full in terms of space, and 78% full in terms of inodes.  One reason for the higher percentage of inodes occupancy is that some users have large numbers of small files. 

 

The support team is currently running reports to help identify files in use by researcher on this partition.  We might reach out to you on Monday to seek your help in helping reduce the inodes use.  You can help by addressing data sets that have large numbers of multiple small files – use zip, bzip2 or tar to create fully usable and accessible single-file archives of these types of data sets _if_ they are not in active use at the moment. 

 

Thanks for your attention to this update, and thanks in advance for any self-maintenance you are able to do to help ensure this critical resource remains available and highly functional for everyone.  Please do not hesitate to contact me directly if you have any questions, or reach out to [log in to unmask] if you need help with anything.

 

Best regards,

Rajiv

 

 

 

Rajiv Shridhar

Director – Systems & Production Services, Information Technology Services

Northeastern University

360 Huntington Avenue, 302-216, Boston, MA 02115

(617) 373-7547 | [log in to unmask]

 



To unsubscribe from the DISCOVERY list, click the following link:
https://listserv.neu.edu/cgi-bin/wa?SUBED1=DISCOVERY&A=1