Grid Monitoring For ZEUS

1
Grid MonitoringFor ZEUS
Alexander Chernyack
DESY Hamburg 14-Sep-07 Summer students seminar
1
2
Outline

• Introduction to Grid
• The Enabling Grids for E-sciencE
• ZEUS on the GRID
• Site Availability Monitoring (SAM)
• SAM for ZEUS
• Summary

2
3
What is Grid?

• A computational grid is a hardware and software
  infrastructure that provides dependable,
  consistent, pervasive, and inexpensive access to
  high-end computational capabilities.
• Carl Kesselman and Ian Foster (1998 )

• Grid is a system that
• coordinates computing resources that are not
  subject to centralized control
• uses standard, open, general-purpose protocols
  and interfaces
• provides basic computational services to the end
  user

3
4
The Enabling Grids for E-sciencE

• 36,000 CPU
• 5 PB disk (5 million Gigabytes)
• 30,000 concurrent jobs

• About 20 projects
• ZEUS, ALICE, ATLAS
• GATE, GPS_at_, CDSS
• Mammogrid, etc.

Project brings together scientists and engineers
from more than 240 institutions in 45 countries
5
ZEUS on the Grid
6
ZEUS on the Grid

• Monte Carlo Production
• MOZART - detector simulation
• CZAR, ZGANA, TLT-ZGANA - trigger simulation
• ZEPHYR - reconstruction program

7
Statistic of completed jobs
Some inefficiencies caused by problems at grid
sites
Solution
Active monitoring of grid services
8
Grid components
CE
Minimal site configuration

• UI User Interface
• user entry point to GRID
• RB Resource Broker
• CE computing element
• batch farm server
• WN Worker Node
• runs the user program
• SE Storage Element
• RLS Replica Location Service
• Replica Metadata Catalog, Local Replica Catalog,
  Replica Manager
• IS information service

Core at DESY
RLS
RB
IS
N x UI
N x UI
N x UI
8
9
Service Availability Monitor

• SAM aims to provide a site independent,
  centralized and uniform monitoring tool for all
  grid services

• Run different sensors on sites (e.g. SE, CE,
  testjob)
• Publish results to DB and web page

10
SAM for ZEUS

• I have implemented for the first time SAM for the
  ZEUS experiment
• We can now run standard SAM tests (sensors)
• and have developed ZEUS specific tests
• These tests have been registered in a centrally
  maintained DB (at CERN)
• They are executed regularly on all Grid sites
  supporting ZEUS VO
• The results can be accessed via web

11
ZEUS Tests

• Aim test all aspects of grid services required
  by MC jobs
• Assign error status to unsuitable sites

What we test

• Storage Element (SE)
• Computing Element (CE)
• Worker Nodes (WN)
• File Catalog
• Access to frequently used files

12
ZEUS Tests details

• Storage Element (SE)
• registering new files, transfer between sites,
  replication and removal
• Computing Element (CE)
• job submission and scheduling
• Worker Nodes (WN)
• Local disk access, proper environment
• File Catalog
• Accessing files via logical name
• Access to frequently used files
• Geometry and calibration constants (GAFs)
• Scripts, configuration files
• executables

13
Test CE-sft-zeusmc-files

• Check availability of ZEUS files needed for
  running MC jobs.
• Copy these files to WN using ZEUS toolkit.

14
Summary

• Done
• SAM has been deployed .
• A new sensors has been written to test ZEUS
  specific functionality needed for MC production.
• Next step
• Finish a simple MC test job with real physics.
• Automatic exclusion of faulty sites

14