How to install Lustre on Ubuntu 9.04 « Ahmet Uludağ’ın günlüğü. Lustre is a popular cluster file system. You can check out here for more information. Installing lustre is somewhat involved on some operating systems. As to ubuntu, there are a few instructions and they may not work for last versions. So, I wrote this article to ease it for you. Although, I tested this installation on Ubuntu 9.04 (Jaunty Jackalope), it should work on many other versions as well. - Firstly, we type sudo -s and then root password to use shell as root. - We get some utilities for compiling kernel: apt-get install build-essential bin86 kernel-package libqt3-headers libqt3-mt-dev wget libncurses5 libncurses5-dev - While compiling kernel with lustre patches, new versions of gcc cause some problems. Apt-get install gcc-3.4 mv /usr/bin/gcc /usr/bin/gcc.bak ln -s /usr/bin/gcc-3.4 /usr/bin/gcc We installed gcc and created a symbolic link to use.
. - We install lustre source, lustre patch and lustre utilities: apt-get install lustre-source linux-patch-lustre lustre-utils cd /usr/src reboot. Ubuntu Hardy - How To Build Lustre and its own Kernel (english v. First of All, we have to create our own kernel.The last version of Lustre, supports only kernel 2.6.22.19, and that`s the one we will use here! 1) Install Some basic packages:apt-get install quilt gcc kernel-package libncurses5-dev fakeroot bzip2 2) Download Lustre v1.6.5.1 [1] and Linux Kernel v2.6.22.19 [2] [1] wget [2] wget 3) Unzip these files to /usr/src bunzip2 (linux source) tar -xf (linux source) tar -xf (lustre source) 4) Create some symlinks for the sources (just to make it ease) ln -s (linux source path) ln -s (lustre source path) 6) Redirect SH (to use some new features) rm -f /bin/sh ln -s /bin/bash /bin/sh 8) make-kpkg clean 10) Let`s rock...
Voila!!! [Lustre-discuss] iptables rules for lustre 1.6.x and MGS recover. Using Pacemaker with Lustre. From Lustre.org (Updated: Jan 2010) This content was submitted by an external contributor. We provide this information as a resource for the Lustre™ open-source community, but we make no representation as to the accuracy, completeness or reliability of this information.
This page describes how to configure and use Pacemaker with Lustre failover. For more about Lustre failover, see Configuring Lustre for Failover. Setting Up Cluster Communications Communication between the nodes of the cluster allows all nodes to “see” each other. An introduction to the setup, configuration and operation of a Pacemaker cluster can be found in: Setting Up the corosync Communication Stack The corosync communication stack, developed as part of the OpenAIS project, supports all the communication needs of the cluster. Note: If corosync is not included in your distribution, your distribution may include the complete OpenAIS package. Complete the following steps to set up the corosync communication stack: 1. 2. 3. 4. A Blog - The Tinkergeek Blog. Kernel + Debian. Lustre ports iptables.
LustrePlusXen < Main < TWiki. Configuring the Lustre File System. From Lustre.org (Updated: Sep 2009) This page describes how to configure a simple Lustre™ file system comprised of a combined MGS/MDT, an OST and a client. The administrative utilities provided with Lustre, however, can be used to set up systems with many different configurations. Note: We recommend that you use dotted-quad (dot-decimal) notation for IP addresses (IPv4) rather than host names. This aids in reading debug logs and helps when debugging configurations with multiple interfaces.
Configuring the Lustre File System This section contains a procedure for configuring the Lustre File System. To configure Lustre Networking (LNET) and the Lustre file system, complete these steps: 1. Options lnet networks=<network interfaces that LNET can use> This step restricts LNET to using only the specified network interfaces and prevents LNET from using all network interfaces. 2. 3.
Mkfs.lustre --fsname=<fsname> --mgs --mdt <block device name> The default file system name (fsname) is lustre. 4. 5. 6. 7. Lustre Configuration Example. From Lustre.org (Updated: Sep 2009) This Lustre™ configuration example illustrates the configuration steps for a simple Lustre installation comprising a combined MGS/MDT, an OST and a client, where: 1. Define the module options for Lustre networking (LNET), by adding this line to the /etc/modprobe.conf file. options lnet networks=tcp 2.
Create a combined MGS/MDT file system on the block device. [root@mds /]# mkfs.lustre --fsname=temp --mgs --mdt /dev/sdb This command generates this output: 3. [root@mds /]# mount -t lustre /dev/sdb /mnt/mdt Lustre: temp-MDT0000: new disk, initializing Lustre: 3009:0:(lproc_mds.c:262:lprocfs_wr_group_upcall()) \ temp-MDT0000: group upcall set to /usr/sbin/l_getgroups Lustre: temp-MDT0000.mdt: set parameter \ group_upcall=/usr/sbin/l_getgroups Lustre: Server temp-MDT0000 on device /dev/sdb has started 4.
[root@oss1 /]# mkfs.lustre --ost --fsname=temp --mgsnode= 10.2.0.1@tcp0 /dev/sdb The command generates this output: 5. Shortly afterwards, this output appears: 6. Improving performance of small files on Lustre : Atul's Blog abo. Everyone knows that Lustre filesystem was designed for large sequential IO type of workload and it works beautifully for such IO pattern. Until few years back HPC sites used to have Lustre only as their scratch filesystem. But recently, more and more sites are rolling out Lustre as home, general purpose data and scratch filesystem. Just have a look at: Using Lustre for such diverse workload presents interesting challenges especially when used for home directories.
User home directories usually have tons of small files and access pattern is random. Lustre performance differs significantly in home directory usage compared to scratch space. 1. By deafult, Lustre 1.8.1.1 will have multiple type of debug messages recorded by every client. Downside of disabling debug messages is that, you will have no clue if something goes wrong. 2. By default, Lustre 1.8.1.1 will have 32 MB of dirty cache per OST. 3. By default, Lustre 1.8.1.1 will have 8 RPCs in flight. 4. 5. 6. 7.