Memo on Hall D Priority on the JLab Batch Farm

From GlueXWiki
Jump to: navigation, search
Date: Mon, 28 Sep 2009 11:11:27 -0400
From: "Mark M. Ito" <marki@jlab.org>
MIME-Version: 1.0
To: Dennis Weygand <weygand@jlab.org>, Sandy Philpott <philpott@jlab.org>,
        Graham Heyes <heyes@jlab.org>
Subject: short-term high priority for Hall D

Dennis, Sandy, and Graham,

Just wanted to summarize the discussions I had with each of you last 
Friday concerning farm priority and Hall D.

The conversation was motivated by several things related to Hall D:

(a) need for large share of the farm to address detector design issues
(b) need only arises sporadically (once every few weeks) and only lasts 
for a few days
(c) the current 4% allotment is often not sufficient to have timely 
resolution of these issues
(d) resulting average use is much less than 4%

Talking to Dennis he agreed that Hall D should be able to ask for 
increased priority for a short period of time and have that priority 
reset to the nominal value after the time expires. Consistent with our 
understanding, let me quantify a few of these terms:

increased priority = 20%
short period of time = one week
nominal value = 4.4%

Talking to Sandy, we agreed an appropriate procedure for invoking 
Hall-D-high-priority mode would be to:

1. submit a CCPR
2. send email announcing the request to Dennis, Graham and Sandy

Dennis will able to monitor the situation using the standard tools.

Is there anything that we need to settle before I announce this option 
to my colleagues?

  -- Mark