Agenda and Minutes, 5 December 2007
Agenda: Wed, 5 December 2007
Minutes
Attendance: Adam Brian Scott Patrick Stuart Kent Ed Felippo Keith T Keith R Warren Adam Mercer John Z Announcements: Discussion of draft report from ad-hoc Committee: Keith Riles gave an overview of how the draft report came to be. He reported that the ad-hoc committee was asked by the Direcorate to comment on control room software including DTT, Matlab, and tools. He reported that the committee went beyond its charge because of the general consensus that Patrick summarized his points as developed in the notes he circulated ... Stuart reminded people that calibration was necessary in real time ... Duncan mentioned his thoughts that the division between CDS and DASWG has often led to weaknesses in analysis ... Keith Thorne thinks that support people are really needed. In anything running on-line, it needs to be able to deal with gaps inside the data ... John said that DMT is robust and can be used online or offline and has the mechanisms available ... Greg pointed out that the DMT uses the frame-broadcaster to pick up the data. Keith added that the DMT is looking at the state-vector. Kipp mentioned that we don't yet have an interface between different analyses. John mentioned that triggers are being written to the databases. Patrick asked what about the interactions that may arise when jobs want to access data very quickly and in large amounts. Keith agreed that having h(t) online will be important. The issue of latency was discussed in much detail again. Patrick summarized: the required latency will dictate the requirements on software. There was consensus about being able to discover and locate data quickly. Ed reminded people that the h(t) generation could live on a different server and provide a different service. Duncan thinks there is a danger of going into too much detail. There was a lot of discussion of the nature of online searches and their latency. People agreed that having some early warning searches would be good ..... John, Kipp, Patrick talked about ..... Stuart asked what the amount of data Patrick will post minutes and seek more comments .... MatApps: Keith has been doing work on the new version of MatLab. He asked if there were any problems with Matlab on CentOS 5.0 and running elsewhere. DMT: Ed and John released FrameCPP and DOL are compilable on CentOS. Adam will start to build these rpms ....... he has built the framecpp and fc4 and centos5. It is in testing on both archives. Also has DOL, will run test and then put into testing repo. Glue: No updates on the software side. Duncan asked it the replication issues have been resolved. No report; Igor did announce an LSCsegFind1 for testing? Patrick will ask Igor to send around a report. Duncan reported that Deelman is still interested in using Glue for workflow development. They want to develop a workflow development package as part of Pegasus. Patrick said keep moving forward on this. Stuart asked if Ben's stuff for publishing is available in a released glue? Duncan said much of that is in there already. Duncan will double check that everything is in. Stuart would like to have glue be official scripts for publishing. Kipp moved over itertools and order log(n) in segment list. LAL/LALApps: Patrick said that Jolien is beginnign to think about having a workshop to figure out where to go next. Warren volunteered to help organize that. LDAS: Ed reported that the port to CentOS 5.0 seems to be in place. LDG: Greg made a slight change to the condor configuration change. He will send an e-mail about this. There were two variables that needed to be set for 4.4 and needed to be set for 4.5 when the server is installed. This is more than a change in the documentation. Patrick asked about the version number issue that arises from this? Greg suggested adding a minor version number increment in the file on the system and add an ldg-version. Felippo is working with Globus MDS to for monitoring. He is still trying to configure webMDS .... LDM: Felippo had nothing to report. LDR: things are going to Cardiff. LIGOtools: no report. MatApps: no report. Metaio: Duncan is putting together a new release. He will send an e-mail to Adam to put into lscsoft. Onasys: no report. OSG-LIGO: Britta is making progress to run on LDG; can now run on LHO. Still having problems with LLO and Nemo. Working to get E@H on OSG. Patrick asked about the big picture side of that. Britta has not heard from them about that. Warren has not heard from them either. Kent asked if there was any more experience running with WSGram. Kent mentioned to the 1.5 day site admin meeting in December; Brian will go. AOB: Event Candidate Database and LARS will be discussed next week to start the design process related to these activities. Once we've had a discussion next week, we will have a joint DAC/DASWG meeting to follow up.
$Id: 071205.html,v 1.1 2007/12/19 18:46:18 patrick Exp $