SSS Deployment using OSCAR

1
SSS Deployment using OSCAR
Aug 2004, Argonne, IL

SSS Face-to-face meeting

• John Mugler, Thomas Naughton, Phil Pfeiffer
  Stephen Scott

2
OSCAR Cluster Toolkit

• Framework for cluster management
• simplifies installation, configuration and
  operation
• reduces time/learning curve for cluster build
• requires pre-installed headnode w. supported
  Linux distribution
• thereafter wizard guides user thru setup/install
  of entire cluster
• Package-based framework
• Content Software Configuration, Tests, Docs
• Types
• Core SIS, C3, Switcher, ODA, OPD, (Support
  Libs)
• Non-core selected third-party
• Access repositories accessible via OPD/OPDer

3
OSCAR Wizard
OSCAR-3.0 release
4
Using OSCAR for SSS

• Problem Helping users obtain and install SSS
  software.
• Solution Leverage OSCAR framework to package and
  distribute the SSS suite, sss-oscar.
• sss-oscar ? A release of OSCAR containing all
  SSS software in single downloadable bundle.

5
OSCAR-ized SSS Components

• Bamboo Queue/Job Manager
• BLCR Berkeley Checkpoint/Restart
• Gold Accounting Allocation Management System
• LAM/MPI (w/ BLCR) Checkpoint/Restart enabled
  MPI
• MAUI-SSS Job Scheduler
• SSSLib SSS Communication library
• Includes SD, EM, PM, BCM, NSM, NWI
• Warehouse Distributed System Monitor
• MPD2 MPI Process Manager

As of April 2004
6
Current Status

• Several 0.2 cuts (latest 0.2a8)
• 2 items remain in 0.2 Tracker, http//sf.net/proj
  ects/sss-oscar/
• Both items have fixes in CVS, pending testing
• Added OSCAR_PACKAGE_TEST_HOME and work arounds
  for current testing framework
• Ready for 0.2a9 can test during meeting!
• After testing post as pre-release on main www
  site
• Starting work on 0.3, etc.
• New Gold pkg
• OSCAR support for part of BCWG schema

7
TODO

• Integrate Gold into new releases
• Integrate APItest into OSCAR
• SSS Component authors create their APItest cases
• Update individual SSS Pkgs as needed
• Update/Improve Documentation for v1.0
• Start weekly builds for testing (next slide)
• Improve testing/bug reporting (fixing)

8
Release Schedule

• ADJUST - previous Sept. 3 freeze date for SC04
  release.
• Nov 8 SC04 release sss-oscar-1.0
• Whoo-hooo! ?
• Oct 4 SC04 freeze
• No changes except to fix approved bugs
• Sep weekly builds
• Available first day of week by 12 noon
• Untested as-is tarballs for tests/bug
• Each developer to test their component for
  acceptance
• Your pkg any dependent pkgs install properly
• Report/Respond to bugs in Tracker
• Make appropriate fixes in CVS to remedy any errors

9
Resources

• ORNL Test1 cluster
• Full install tests restore of headnode (2
  compute victims)
• Access via ORNL Login Server (examples/info
  pending)
• Must do reservations/coordinate use
• Restart nodes with care no remote power mgmt
• Developer CVS repository
• Hosted at http//sss-oscar.sf.net
• Account requests torc_at_msr.csm.ornl.gov
• OSCAR Homepage
• http//oscar.OpenClusterGroup.org
• Includes HOWTO Create an OSCAR Package document

10
Validation Testing
Working Components and Interfaces (bold) Spr03
Color code
BCWG
PMWG
RMWG
VIWG
N/A