ComputingProjects < HEPProjects

HEPProjects Web>ComputingProjects (2006/06/27, RamiVanguri) (raw view)EditAttach
---+!!Computing Projects

%TOC%

---+ Ground Rules for this Page

Projects listed here should always include the following:
   * Brief description
   * Starting date
   * Primary person responsible plus nominal supervisor in case of students
   * Link to "clean" copy of accomplishments and status
   * Link to "dirty" copy of status, i.e. the electronic logbook for collaborative development

---+ Projects involving students

---++ Projects involving Undergraduates

---+++ Pnfs Consistency Checker
---++++ Brief Description
Develop, deploy, and package a system that checks the integrity of pnfs in a regularly scheduled fashion, 
finds missing as well as corrupted files, and allows for automatic re-download of the missing or corrupt files
using PhEDEx, if possible.
---++++ Starting Date: April 2006
---++++ RamiVanguri supervised by JamesLetts
---++++ PnfsChecker 
---++++ PnfsCheckerElog

---++ Projects involving Masters students (mostly CS)
---+++ Squid Scalability Investigation
---++++ Brief Description
Develop a set of tests to understand squid scalability, and use them to test squid in Xen, and compare
it with squid deployed on normal hardware.
---++++ Starting Date: April 2006
---++++ KiranKalyan supervised by AbhishekSinghRana
---++++ SquidTests
---++++ SquidTestsElog


---++ Projects involving Ph.D. Graduate students

---+++ On demand Job Monitoring
---++++ Brief Description
Develop an on demand job monitoring tool that allows users to do ps, ls, top, less, etc. on
their jobs running on the grid without having to know where they run.
---++++ Completed: Fall 2005
---++++ ShihChiehHsu supervised by ElliotLipeles
---++++ JobMon

---+++ Generic Connection Broker Scalability Tests
---+++++ Summary
The Condor Generic Connection Broker (GCB) is a piece of software developed independently for Condor to allow Condor daemons to run
over a firewall or through a NAT without difficulty.  Scalibility was tested both in local deployment, and in Beta deployment on the OSG.  Currently it is used by both the NamCAF and the OSGCAF.

Completed: Spring 2006
MatthewNorman supervised by Igor Sfiligoi (FNAL)

---+++ Condor-C Scalability Tests
---+++++ Summary
Condor-C is Condor's distributed schedd approach to computing, involving deploying schedulers on multiple nodes to reduce load.  Scaling 
tests were done with an eye toward deployment at FNAL, but concluded that Condor-C does not scale with kerberos, and that switching to 
GSI was currently impractical.

Completed: December 2005
MatthewNorman supervised by Igor Sfiligoi (FNAL)
---+++++ Resources:
[[http://www-cdf.fnal.gov/htbin/twiki/bin/view/Main/MatthewNormanCondorC][CDF Condor-C guide]]
[[http://www.cs.wisc.edu/condor/manual/v6.7/5_3Grid_Universe.html][Condor Manual Page for Grid Computing]]

---+++ Condor Scalability Tests
---+++++ Summary
To reduce scheduler load, tests were done with deploying multiple schedds on a single node to reduce the single scheduler bottleneck.  The 
first tests failed, but a second set using web caching to offload the burden of user tarball transfer were more successful.  The system was 
declared satisfactory and was deployed in Spring 2006 at FNAL.

Completed: Spring 2006
MatthewNorman supervised by ElliotLipeles and Igor Sfiligoi (FNAL)
---+++++ Resources:
[[http://www-cdf.fnal.gov/htbin/twiki/bin/view/Main/MultipleScheddEnvironment][Old Multiple Schedd Page]]


---+++ OSG CAF
---+++++ Brief Description
OSG-CAF was to be a CDF gateway onto the OSG, allowing single-point-of-submission access to a glide-in based Condor pool that could 
harvest opportunistic resources from the entire Grid.  Buzzwords aside, it lets CDF users treat the OSG as a giant CAF, and keeps the 
headnodes and other maintenance tasks centralized at FNAL.  Later split into two functions
---+++++ Starting Date: January 2006
---+++++ MatthewNorman supervised by Igor Sfiligoi (FNAL)
---+++++ Current Status: Delayed due to lack of interest (OSGCAF) or deployed (NamCAF)
---+++++ Resources:
[[http://dcafmon.fnal.gov/osgcaf/][OSG Monitoring Page]]

---+ Projects Involving Staff

---+++ NFS Lite Compute Element for OSG
---++++ Brief Description
Develop a configuration for an OSG 0.4.1 Compute Element that does not require any filesystem exports.
---++++ Completed April 2006
---++++ TerrenceMartin
---++++ NfsLiteComputeElement

---+++ Performance testing of NfsLiteComputeElement
---++++ Brief Description
---++++ Starting Date: June 2006
---++++ HaifengPi
---++++ NfsLiteComputeElementPerformance
---++++ NfsLiteComputeElementPerformanceElog
Topic revision: r1 - 2006/06/27 - 17:49:59 - RamiVanguri
HEPProjects
Copyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback