Summary PHYSGI Meeting 8/1/94

B. Gibbard

Next Meeting:

Production User Input for PHYSGI01:

1. Only E852 was represented at this week's meeting.
2. During the past week the major CPU intensive user was E881 (Hackenburg et al) with some production activity by E850. There continues to be substantial tape copying activity on PHYSGI01, primarily by E852.
3. E852 will begin to perform intermittent calibration oriented production processing, primarily on PHYSGI03, in preparation for a massive analysis pass through the data from the recent AGS run. This analysis pass is expected to begin not later than October 1.
4. E850 is expected to continue production processing on PHYSGI03 this week.
5. E881 is expected to continue its use of PHYSGI01 and PHYSGI03. It was recently estimated that they were 3/4 done with their data analysis pass.

Plan for week beginning 8/1/94:

• 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 currently allowed to run up to 6 concurrent "production" processes per computer (PHYSGI01 or PHYSGI03). On each computer only one of these 6 processes may be run at full priority, (100).
• All "production" processes beyond the first are to be niced to 2 (102). This allows for the possibility of assigning medium priority jobs, niced to 1 (101), at some point in the future.
• 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.
• The tape stackers represent a relatively scarce resource. It is hoped that negotiation between users will result in mutually acceptable sharing. If users find that not to be the case, rules will be established to guarantee sharing which is as equitable as possible, consistent with the overall mission of the PHYSGI computing system.

Other issues:

1. The new challenge machine intended to replace PHYSGI01 is now in operation as an Ethernet connected compute server, PHYSGI03. At some point, currently estimated to be about 2 week from now, all peripherals and interfaces currently on PHYSGI01 will be move to the Challenge machine. It will be renamed PHYSGI01 and will assume all roles currently filled by the old PHYSGI01. The old PHYSGI01 will be retired to private practice. It is expected that this will required a minimum of one full day of down time which will likely be followed by a period of instability as various unexpected problems are sorted out.
2. There have been numerous reports of network problems associated with the use of PHYSGI01 as well as other Physics Department machines resently. CCD believes that Charybdis, the router serving Physics, is heavily overloaded. An upgrade of this router is now expected to occur and it is hoped that this will substantially relieve these recently observed problems.
3. The interaction of private multi-cast tunnels with MBone (desk top video conferencing) tunnels has the potential for substantially increasing network traffic. In the absence of an important need for such tunnels, their use is discouraged on subnets in the Physics Department.
4. At this time there are no known problems with the 8mm tape stackers.

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.