Go to previous page / next page of Tier3 site log

19. 12. 2012 Largest directories via Chimera

New version

Simply consult the daily report:

• http://t3mon.psi.ch/ganglia/PSIT3-custom/v_pnfs_top_dirs.txt
• http://t3mon.psi.ch/ganglia/PSIT3-custom/v_pnfs_top_dirs.html

Outdated since May 2014

Based on Fabio's monitoring function the following function & query can be used to extract the largest folders in a given PNFS directory similar to the previous space usage reporting.

CREATE OR REPLACE FUNCTION pnfs_subdirs_du_r(VARCHAR,VARCHAR,INTEGER) RETURNS SETOF v_pnfs AS $$
DECLARE
    fileordir1   v_pnfs%rowtype;
    fileordir2   v_pnfs%rowtype;
    RESULT       BIGINT;
BEGIN
     
    FOR fileordir1 IN ( SELECT * FROM pnfs_dir($1) WHERE path LIKE '/'||$1||'/%' AND depth < $3 ) loop
     
       fileordir2 = fileordir1;
       SELECT INTO RESULT * FROM pnfs_dir_du( SUBSTRING ( fileordir1.path , 2 , LENGTH(fileordir1.path)) , $2) ;
       fileordir2.isize = RESULT;
       fileordir2.depth = fileordir2.depth + 1;
       RETURN NEXT fileordir2;
     
    END loop;
END;
$$ LANGUAGE plpgsql STABLE;

Then use the following query to get all folders below a given folder (up to depth D) ordered in decreasing order by size: Don't use that query on very large folders with many subfolders!

SELECT * FROM pnfs_subdirs_du_r('pnfs/psi.ch/cms/trivcat/store/user/dmeister','MB',D) WHERE isize IS NOT NULL ORDER BY isize DESC LIMIT 100;

Of course there would be much more efficient methods to do this; but it works!

-- DanielMeister - 2012-12-19

---

Go to previous page / next page of Tier3 site log