Admininstering Lustre Based Clusters

Trainingsziel

Dieser Kurs vermittelt folgende Kenntnisse und Fähigkeiten:

• Describe the Lustre environment



• Understand and illustrate the Lustre cluster architecture



• Perform installation and initial configuration activities



• Troubleshoot and resolve typical operational issues



• Explain and perform advanced profiling activities

Teilnehmerkreis

Students who can benefit from this course include system administrations, architects, and integrators.

Voraussetzungen

Um diesen Kurs erfolgreich absolvieren zu können, sollten Sie über folgende Vorkenntnisse verfügen:

• Students must have Linux System Administration experience, be comfortable compiling the Linux kernel and able to set up networking.

Überblick über den Inhalt des Trainings

Module 1 - Introduction to Lustre

* What is Lustre?

* Explore Lustre architecture

* List the Lustre features

* Describe Lustre software modules

* Describe Lustre networing

* Exercise - Understanding Lustre

Module 2 - Architecting Lustre Clusters

* Examine Lustre requirements

* List systems hardware requirements

* List Lustre node count and size

* Describe Lustre Networking options

* Describe storage and throughput

* Exercise - Cluster Architecture

Module 3 - Lustre Configuration - RPM

* Describe Lustre releases

* Examine required download information

* Describe Patchless Client information

* Install Lustre software

* Identify Lustre configuration principles

* Lab - Install Lustre software using RPM

Module 4 - LNET - Lustre Communications

* Understanding Lustre Network

* Basic Configuration

* Introduction to routing

* Lab - Test Connectivity

Module 5 - Building a Lustre Cluster

* Simple TCP Network

* Lustre with Combined MGS/MDT Node

* Lustre with Separate MGS and MDT Nodes

* Multi-homed Servers

* Failover Concepts

* MDS and OST failover

* Lustre and Quotas

* Lab

Module 6 - The Lustre Client

* POSIX Compliance

* Introduction to Locking

* Client interactions

* Lab - Multi-client interactions

Module 7 - Advance Lustre Client

* Lustre Stiping and I/O Options

* Lustre and ACL''''s

* Introduction to possible Client issues

* Lab - Quota exercise

Module 8 - Lustre HA Clustering

* High Availability Discussion

* Describe array cache and failover

* Describe disk replication

* Lab - Linux-HA setup and testing

Module 9 - Advanced LNET

* Multi-homed hosts

* Bonding and multi-rail

* Routing

* Testing

* Lab - Troubleshooting

Module 10 - Introduction to Troubleshooting

* Describe troubleshooting methodology

* Describe general actions

* Generate debuf information

* Identify stuck modules

* Lab - Troubleshooting

Module 11 - Advanced Troubleshooting

* Manage full OST''''s

* Manage failed OST''''s

* Perform recovery

* Troubleshoot storage failures

* Interpret error messages

* Lab - Troubleshooting