[Title] [Prev] [Contents] [Next] [End]


Contents


Figures

Tables

Preface

Audience
Organization
Related Documents
Technical Assistance

1. Before Installation

LSF Packaging and Licensing
System Requirements
Hardware and Software Requirements
Disk Space Requirements
Obtaining a Software License
Choosing Hosts for the Cluster
LSF Client and Server Hosts
Planning the Installation
Choose Components
Mixed Clusters
Naming the LSF Cluster
Choosing the Primary LSF Administrator
Choosing the Installation Directories
Directory Structure for Mixed Clusters
Directory Structure
Backing Up the System Disk

2. Installation

Overview
Privileges Needed for Installation
Reading the Distribution Medium
Loading LSF From a Tape
Uncompressing a Compressed Tar File
Getting a License Key
Exporting the /usr/local/lsf/mnt Directory
Installing LSF with lsfsetup
Installing the First Host Type
Installing Each Additional Host Type
Additional Steps on Each LSF Server Host
Installing on AFS
Installing the LSF AFS Distribution
AFS Token Encryption
AFS Token Renewal Kit
Installing on DCE/DFS
Installing the LSF DCE Distribution
Credential Encryption
Upgrading From a Previous Version
Registering LSF Service Ports
NIS Services Database
Configuring Services in lsf.conf
Registering NQS Service
Starting LSF Servers at Boot Time
Licensing the Different LSF Components
Getting License Key Information
Setting Up the License Key
Installing a DEMO License
Installing a New Permanent License
Starting the License Server Daemons at Boot Time
Adding a Permanent License

3. Configuring LSF Cluster

Initial Configuration
Setting Up LSF Client Hosts
Checking the LSF Configuration
Testing the LSF Cluster
Testing LIM
Testing RES
Testing LSF Batch
Configuring LSF MultiCluster
Configuring LSF JobScheduler
Providing LSF to Users
Using xlsadmin

4. LSF Concepts

Definitions
LSF Product Suite
Jobs, Tasks, and Commands
Hosts, Machines, and Computers
Clusters
Local and Remote Hosts
Submission, Master, and Execution Hosts
Fault Tolerance
Shared Directories and Files
Shared User Directories
Executables and the PATH Environment Variable
Using LSF without Shared File Systems
Resource Requirements
Host Naming
Hosts with Multiple Addresses
Remote Execution Control
User Authentication Methods
Authentication using Privileged Ports
Authentication using Identification Daemons
External Authentication
Security of LSF Authentication
How LSF Chooses Authentication Methods
Remote Execution Permission
User Account Mapping
Job Starter
Load Sharing with LSF Base
Time Windows
How LSF Batch Schedules Jobs
Job States
Eligible Hosts
Dispatch Windows
Run Windows
Resource Requirements
Host Lists
Host Load Levels
Order of Job Dispatching
Job Slot Limits
User Job Slot Limits
Host Job Slot Limits
Queue Job Slot Limits
Resource Limits and Resource Usage
Scheduling Policy
Suspending Jobs
Resuming Suspended Jobs
User Suspended Jobs
Interactive Batch Job Support
Pre- and Post-execution Commands
Checkpointing and Migration
Job Migration
Job Control Actions
Resource Reservation
Processor Reservation
Remote File Access
Job Requeue
External Submission and Execution Executables
External Events and eeventd
External Load Indices and ELIM

5. Managing LSF Base

Managing Error Logs
LSF Daemon Error Log
FLEXlm Log
Controlling LIM and RES Daemons
Checking Host Status
Restarting LIM and RES
Remote Startup of LIM and RES
Shutting down LIM and RES
Locking and Unlocking Hosts
Managing LSF Configuration
Overview of LSF Configuration Files
Configuration File Formats
Example Configuration Files
Changing LIM Configuration
Reconfiguring an LSF Cluster
External Load Indices
Writing an External LIM
Configuring External Load Indices
Overriding Built-In Load Indices
LIM Policies
Tuning CPU Factors
Tuning LIM Load Thresholds
Cluster Monitoring with LSF
LSF License Management
How FLEXlm Works
Updating an LSF License
Changing the FLEXlm Server TCP Port
Modifying LSF Components and Licensing

6. Managing LSF Batch

Managing LSF Batch Logs
LSF Batch Accounting Log
LSF Batch Event Log
Controlling LSF Batch Servers
LSF Batch System Status
Remote Start-up of sbatchd
Restarting sbatchd
Shutting Down LSF Batch Daemons
Opening and Closing of Batch Server Hosts
Controlling LSF Batch Queues
bqueues - Queue Status
Opening and Closing Queues
Activating and Inactivating Queues
Managing LSF Batch Configuration
Adding a Batch Server Host
Removing a Batch Server Host
Adding a Batch Queue
Removing a Batch Queue
Controlling LSF Batch Jobs
Moving Jobs - bswitch, btop, and bbot
Signalling Jobs - bstop, bresume, and bkill
Tuning LSF Batch
Controlling Interference via Load Conditions
Understanding Suspended Jobs
Controlling Fairshare
Dispatch and Run Windows
Controlling Job Slot Limits
Resource Reservation
Processor Reservation
Controlling Job Execution Environment
Understanding Job Execution Environment
NICE Value
Resource Limits
Pre-execution and Post-execution commands
Using A Job Starter
Using Licensed Software with LSF Batch
Host Locked Licenses
Host Locked Counted Licenses
Floating Licenses
Example LSF Batch Configuration Files
Example Queues
Example lsb.hosts file
Managing LSF Cluster Using xlsadmin

7. Managing LSF JobScheduler

System Calendars
External Event Management
The EEVENTD Protocol
File Event Handling

8. Managing LSF MultiCluster

What is LSF MultiCluster?
Creating a New Cluster
The lsf.shared File
The lsf.cluster.cluster File
Root Access
LSF Batch Configuration
Inter-cluster Load and Host Information Sharing
Running Interactive Jobs on Remote Clusters
Distributing Batch Jobs across Clusters
Account Mapping Between Clusters
User Level Account Mapping

9. LSF Base Configuration Reference

The lsf.conf File
The lsf.shared File
Clusters
Host Types
Host Models
Resources
External Load Indices
The lsf.cluster.cluster File
Parameters
LSF Administrators
Hosts
The lsf.task and lsf.task.cluster Files
Local Tasks
Remote Tasks
The hosts File
The lsf.sudoers File

10. LSF Batch Configuration Reference

The lsb.params File
Parameters
Handling Cray NQS Incompatibilities
The lsb.users File
UNIX User Groups
LSF Batch User Groups
User and Group Job Slot Limits
The lsb.hosts File
Host Section
Host Groups
Host Partitions
The lsb.queues File
General Parameters
Processor Reservation for Parallel Jobs
Flexible Expressions for Queue Scheduling
Load Thresholds
Resource Limits
Eligible Hosts and Users
Scheduling Policy
Migration
Queue-Level Pre-/Post-Execution Commands
Job Starter
Configurable Job Control Actions
Automatic Job Requeue
Exclusive Job Requeue
Default Host Specification for CPU Speed Scaling
NQS Forward Queues
The lsb.nqsmaps File
Hosts
Users

A. Troubleshooting and Error Messages

Error Log Messages
Finding the Error Logs
Shared File Access
Common LSF Problems
LIM Dies Quietly
LIM Unavailable
RES Does Not Start
User Permission Denied
Non-uniform File Name Space
Common LSF Batch Problems
Batch Daemons Die Quietly
sbatchd Starts But mbatchd Does Not
sbatchd: unknown service
Host Not Used By LSF Batch
Error Messages
General Errors
Configuration Errors
LIM Messages
RES Messages
LSF Batch Messages

B. Sample Installations

Install for the First Time
Getting License Key Information
Installing a Permanent License
Setting Up a Host

C. LSF Directories

D. Sample System Support

IRIX 6 Processor Sets
Time-Based Processor Allocation
User-Based Processor Allocation
Other Situations
IBM SP-2 Support
Support for HP Exemplar Technical Servers
Adding Load Indices Definitions
Adding Queue Definitions
Configuring NQS Interoperation
Registering LSF with NQS
lsb.nqsmaps
Configuring Queues for NQS jobs
Handling Cray NQS Incompatibilities
Support for Atria ClearCase

E. LSF on Windows NT

Requirements
Recommended
Features and Limitations
Installation
Starting LSF Service
License Management
Demo License
Permanent License
Mail
Environment Variable Handling
Windows NT 4.0
Security Issues
Heterogeneous NT/UNIX Environments
Differences between LSF for UNIX and NT
Miscellaneous

F. New Features in LSF 3.0

Windows NT and Additional Unix Platform Support
Interactive Jobs with Batch Scheduling Control
Job Level Resource Usage
Enhanced Resource Limit Control
Resource Reservation
Processor Reservation For Parallel Jobs
Flexible Expressions for Queue Scheduling
Host Preferences
Generalized Checkpointing Support
Job Starter
Configurable Job Control Actions
Unlimited Number of Load Indices and Resources
Enhanced Preemptive Scheduling
Per-Host Job Slot Limit of a Queue
Remote Startup
Exclusive Job Requeue
LSF MultiCluster
File Status Events
External Events
System Calendars


[Title] [Prev] [Contents] [Next] [End]

doc@platform.com

Copyright © 1994-1997 Platform Computing Corporation.
All rights reserved.