Summary PHYSGI Meeting 6/6/94

B. Gibbard

Next Meeting:

The next meeting will be Monday, June 13, 11:30 - 12:00, in the Orange Room.

Production Users of PHYSGI01:

During the past week the largest CPU intensive user was E818 (Hackenburg et al.) which ran 6 CPU intensive jobs niced to a priority of 102. These jobs clearly yield time to jobs at priority 100 and 101 but still managed to consume the bulk of the PHYSGI time due to the generally low level of usage by other experiments. E850 (Gushue et al.) have temporarily moved their CPU activity to a new Challenge server in the Chemistry Dept. The success of E818's running at reduced priority has prompted a change of the rules for production processing, see below.
Tape activity on PHYSGI was high with substantial contention between users, prompting a statement of policy regarding tape usage and a modification of the TAPMAN routine, see below.
Indications are that similar relatively light CPU usage with heavy tape and disk usage is likely for the coming week.

All experiments which anticipate running more than an single occasional production job during a given week are required to attend the scheduling meeting. If that is not possible they should notify Bruce Gibbard (gibbard@bnl.gov) and Tom Throwe (throwe@rsgi02.rhic.gov) of their needs by Email prior to the Monday 11:30 meeting.

Plan for week beginning 6/6/94:

New rules will be in effect for the following week which are intended give a modest advantage to experiments involved in the current AGS run.

A production process is defined as a process which would accrue an hour or more of CPU time in 8 hours of uncontended running.
Each experiment is allowed to run up to 6 concurrent "production" process, only one of which may be run at full priority, (100).
Experiments not actively engaged in the current AGS run must nice all additional "production" process to 4 (104).
Experiments which are actively engaged in the current AGS run may nice one additional "production" process to 2 (102) but are required to nice "production" processes beyond the first two to 4 (104).
Extended use of a tape drive is defined as use which extends for more than 1 hour.
In the presences of demand by other experiments for tape drives, no experiment should make extended use of more than 2 tape drives.
Experiments not actively engaged in the current run should limit their extend use, in the presence of demand by other experiments, to 1 tape drive.

Other issues:

The TAPMAN utility will be upgraded in the near future so that a user will be able to deallocate a tape drive which is empty independent of who the owner is.

There continue to be problems with the 8 mm tape drives on PHYSGI01. There are indications of at least two independent problems. One problem may be related to incompatibilities between Exabyte, TTI, and IRIX5.2 firm/software. Another may be specific to the PHYSGI01 hardware configuration. Work on these problems will continue with high priority.

Ongoing problems with XCLOCK on PHYSGI01 to which SGI does not seem to have a fix have prompted its being disable for the time being.

Users of PHYSGI01 are still strongly encourage to set a mail forward from their working account on PHYSGI01 to an account on a machine where they frequently check their mail so that they can be easily contacted.