PLEASE READ: New Raith Reservation Policy Semi-Final Proposal Going Into Effect 2/10/2004

John Shott shott at snf.stanford.edu
Sun Feb 8 11:19:31 PST 2004


Mark and the rest of the Raith community:

Again, I applaud your efforts to come up with some reservation guidelines that 
will, hopefully, maximize use of the Raith and avoid hogging problems.

I promised that I would see if I could make some changes to Coral after 
soliciting feedback from the rest of the SNF community.  I did receive a 
certain amount of negative feedback related to the initial proposal of pushing 
out the reservation horizon to 14 days and changing the rule so that the start 
time, rather than the end time, has to fall withing the reservation horizon.

The primary fear is that these rules with encourage hogging on other machines 
and could, particularly in the case of the stsetch, result in everything being 
booked 24x14.  The other argument, that does warrant consideration, is the 
fact that things are sufficiently variable in the research environment, 
particularly with unscheduled equipment downtime, that it is risky to open up 
the reservation horizon too far ... it may result in an unacceptably high 
fraction of cancelled or unused reservations.  While I think that some of 
their fears may be unfounded and, if hogging occurs, they need to come up with 
community-based guidelines for those machines, I've decided that it is best to 
make the following compromise in the proposed changes to Coral.

(1) I plan to push the reservation horizon to 10 days ... better than the 
current 7, but less than the initially discussed 14.  However, I see no reason 
that this conflicts seriously with the proposed reservation guidelines ... 
even though it doesn't allow reservations made in days 11-14.

(2)  I do plan to change the rule so that the START of a reservation falls 
within the reservation horizon.  This will, I think, be a big benefit for the 
raith community (and for stsetch) by allowing each reservation to be a usable 
length.  You've already implemented guidelines to avoid hogging ... and I'm 
hopeful that the ststech community will follow your lead.

I've tested these changes in our development environment and set it up so that 
the production Coral servers will restart with this behavior very close to 
12:01 a.m. on Tuesday morning.

Thanks to all of you for your inputs on this matter,

John





More information about the raith mailing list