VDSers, since some jobs seem to be acting weird and not running fully on the blades – I would suggest doing the Cluster Cleaning protocol that is in the DatabasesVDS folder for EACH blade that you want to run on. You would do this before runnning your job.
Check with a mentor or Dr. B before you do this the first time though.
How to Clean a Cluster/Rid it of Gold Jobs
By: Adam Nguyen, Clifford Ho, and Kathryn Pendleton
Someone please publish this paper anybody!
This paper is now copyrighted, if you plagiarize from it, we will find you. And kill you.
****************Commands listed to be entered with description of their functions in parentheses********************
1) login to SSH Secure Shell (Host: ddfe.cm.utexas.edu)
2) ganglia cpu_user (should only be running jobs on blades 9-16)
3) ps -u username (ex. ps -u plkrs777)
(This should show you if you have any processes running)
4) ssh compute-0-X (where X=blade number i.e. 9-16, do not input a zero before single digit clusters!!! Enter the number for computer node that your job is found on, this command will take you to the node )
5) ps -u username (ex. ps -u plkrs777, to see if there are any processes that can be killed from the specific computer node)
6) kill -9 ###### (To kill a process, use this command, where ###### is the PID number on the left hand side for the specific process you are trying to get rid of)
7) ps -u username (ex. ps -u plkrs777, this time, to verify that the process is no longer in the list of the specific computer node)
8) cd / (This changes your directory to the root directory)
9) cd tmp (This takes you to the temporary directory)
10) ls -l (This displays the long names for each file in the temporary directory, determine which pmvd file is yours, ie. has your user id)
11) rm pvmd.#### (This command removes the file with the filename specified, ie in this ex it is pmvd.#### WARNING ONLY DELETE THE pmvd file THAT IS UNDER YOURE USERNAME, DO NOT DELETE SOMEONE ELSES. OR WE WILL FIND YOU AGAIN. AND KILL YOU AGAIN)
12) rm pvml.#### (This removes the license file)
13) ls -l (To verify the deletion of the file from the temporary directory)
14) exit (to leave the computer node)
15) Repeat steps 1-13 for other nodes requiring cleaning
Filed under: Housekeeping, Protocols, Virtual Screening | Tagged: DDFE, gold, libraries, virtual, Virtual Screening | Leave a comment »