The content of this article or any related information is under the Creative Commons license BY, you can republish this content freely but you must mention the author of this article and indicate the URL of this page: https://www.exabyteinformatica.com/tienda/foro/two-node-cluster-with-suse-t1430.html
1 Two Node Cluster with SUSE Linux business Server high Availability Extension DRBD and OCFS2 KVM / XEN Mark Gardner.
2 During this session this is a SUSE Linux consultant’s forum masking optimum practices. Learn the way to create an easy two-node cluster the usage of SUSE Linux business Server and the excessive Availability Extension. Free your workloads from the bonds of actual hardware. Host dissimilar workloads on this cluster. Gain knowledge of various high availability suggestions with VLS and/or OpenAIS/Pacemaker. During this session you're going to find out how to make use of applied sciences equivalent to DRBD, Pacemaker, OCFS2, OpenAIS and Xen. Gain knowledge of logical firm of clustered features, and turn into exposed to several configuration examples.
3 What's a workload? Physical Server actual Server as virtual Host 1 workload per physical server varied workloads per actual server.
4 Instance Workloads Apache utility Servers (Tomcat/JBoss/Glassfish) DNS Database LDAP an entire virtualized visitor working gadget can be handled as a single workload.
5 Outline your desires maintain them simple Two node KVM/Xen virtualization cluster energetic/active disbursed Filesystem Virtualized visitors with are living Migration, automatic failover on node failure SBD STONITH Pacemaker for helping supplies.
6 What are we attempting to accomplish? Consolidation high Availability disaster recovery more desirable Hardware Utilization stronger provider stages.
7 Cluster Intro A cluster is a kind of parallel or distributed systems that: consists of a set of interconnected complete computers. it is used as a single unified computing resource Dr. Gregory Pfister, in the hunt for Clusters (1995).
8 Forms of Clusters five smartly regularly occurring cluster forms HA: excessive Availability Cluster HTC: excessive Throughput Cluster HPC: high performance Cluster VSC: virtual equipment Cluster Grid Computing.
9 Usual Cluster Configuration.
10 a standard Cluster Hardware Router Node 1 Node 2 Node three SAN.
11 A 2 traditional Node Setup Cluster with Replicated Storage Hardware Router Node 1 Node 2 TCP/IP DRBD.
12 Cluster building Blocks available substances and Workloads high Availability Engine operating gadget Setup community and Fencing local and Shared Storage.
13 Cluster building Blocks purchasable resources and Workloads high Availability Engine operating device Setup community and Fencing native and Shared Storage.
14 Shared Storage Shared storage delivery relies upon heavily on your workload profile. Virtual Machines software components Shared Storage varieties Storage area network (SAN) NFS iscsi pursuits DRBD Replicated Storage Partitioning options.
15 Local Storage and File methods native storage suggestions local storage partitioning method File methods (all want CLVM2) OCFS2 GFS LustreFS CODA FS.
16 Cluster building Blocks available substances and Workloads high Availability Engine operating device Setup community and Fencing native and Shared Storage.
17 Network Configuration Two TCP/IP channels Bonded channels network should help multicast recent patch allows for corosync to supports unicast second most essential part of clustering after respectable storage.
18 Network Bonding Modes Mode=0 (stability-rr): round Robin Mode=1 (active-backup): just one energetic Mode=2 (stability-xor): Transmit according to MAC Mode=3 (broadcast): Transmit on all slaves Mode=4 (802.3ad): Dynamic link Aggregation!!! Requires particular change configuration Mode=5 (steadiness-tlb): Transmit Load balance Mode=6 (stability-alb): Adaptive Load stability essentially the most average modes are: 1, 2, 0, & 4 (4 is most advantageous).
19 Fencing Fencing limits and a lot of cases prevents situations that effect in break up brain Node Fencing STONITH XEN guest destruction SBD resource Fencing LUN Reservation Quorum Turning off switch ports.
20 Cluster constructing Blocks attainable materials and Workloads excessive Availability Engine working device Setup network and Fencing local and Shared Storage.
21 Cluster building Blocks obtainable supplies and Workloads high Availability Engine operating equipment Setup network and Fencing native and Shared Storage.
22 SUSE Linux commercial enterprise high Availability Extension HA Stack from 10 to 11 SLES 10 Heartbeat DRBD 0.7 Yast2-HB OCFS2 / EVMS2 SLE HA 11 OCFS2 ordinary FS Unified CLI Pacemaker openais HA GUI Yast2-DRBD SLE HA 11 SP1 Metro-environment Cluster Storage Quorum insurance Samba Cluster more desirable information Replication Cluster Config Synchronization Node recuperation Yast2-Multipath web GUI part of SLES 10 added in SLE HA eleven added in SLE HA eleven SP1.
23 Cluster Engine structure.
24 Cluster constructing Blocks accessible resources and Workloads high Availability Engine working gadget Setup network and Fencing local and Shared Storage.
25 Who Controls What?
26 Aid brokers useful resource agents are the bits of code that bridge the gap between what happens in the cluster, and what a managed useful resource can do or is doing. Useful resource brokers will also be written in any language. Search for Open Cluster Framework (OCF) brokers earlier than using an agent of one more type. Steer clear of legacy Heartbeatv1 brokers, they are best around for migration from ancient types of heartbeat.
27 Cluster building Blocks purchasable elements and Workloads high Availability Engine working gadget Setup community and Fencing native and Shared Storage.
28 Sample Configurations.
29 What the Do end you purpose Get? high Availability clustering solution service availability 24h a day Oracle Cluster File system 2 (OCFS2) Sharing and Scaling records-access with the aid of distinct nodes allotted remote Block device 8 (DRBD) deliver most economical SAN through disk replication via TCP user-pleasant tools.
30 Setup 2 Servers installed with SUSE Linux enterprise Server 11 SP1. Unallocated area on local storage for Xen guests Bonded network Interfaces (surest if setup on diverse switches). This is our first line of protection against cut up brain eventualities excessive Availability Extension installed and patched to newest version.
31 Messaging Layer: Corosync.
32 Setup Corosync. This can be configured by means of the command line by means of configuring /and so forth/corosync/corosync.conf Requires network multicast for communication with nodes. IPV4 required if you desire redundant rings (IPV6 doesn't yet help redundant rings). A contemporary patch allows for unicast messaging conversation as an alternative of multicast.
33 Corosync Setup (I) start YaST2 Set communication Channel select Port 5405 opt for Multicast address make certain this tackle is entertaining, is filtered out through some excessive conclusion switches. opt for Auto Generate Node id permit security Auth Generate authkey file on first node most effective replica authkey file and corosync.conf to all different cluster nodes.
34 Corosync Setup (II).
35 Delivery Corosync Run command rcopenais birth Run command crm_mon you're going to get a display like this for as much as 2-3 minutes while the cluster configures itself for the primary time.
36 Messaging Layer comprehensive verify Corosync.
37 Setup the Pacemaker GUI when you installation pacemaker the hacluster person is created and not using a password 1) Run command: passwd hacluster 2) Run command: crm_gui three) Enter in the password you put in step 1.
38 Pacemaker GUI.
39 Configure the CRM exchange Batch limit from 30 to 1 alternate No Quorum coverage to freeze alternate Stonith motion to poweroff Uncheck Stonith Enabled.
40 Basic CRM Configuration.
41 DRBD Setup.
42 Initialize the Disks investigate file syntax Drbdadm dump all replica the DRBD configuration information to the other node: scp /and so forth/drbd.conf node2:/and so on/ scp /and so on/drbd.d/* node2:/and many others/drbd.d/ Initialize the meta records on each techniques by using coming into here on each and every node. drbdadm -- --ignore-sanity-checks create-md r0 rcdrbd start check with rcdrbd popularity.
43 Initial Synchronization delivery the resync procedure to your intended simple node (node1 in this case): drbdadm -- --overwrite-statistics-of-peer basic r0 investigate the fame once again with rcdrbd repute and also you get:... m:res cs ro ds p set up fstype 0:r0 linked simple/Secondary UpToDate/UpToDate C The repute in the ds row should be UpToDate/UpToDate Set node1 as primary node: drbdadm fundamental r0.
44 Add DRBD to Pacemaker primitive drbd-r0 ocf:linbit:drbd \ params drbd_resource="r0" \ op video display interval="30" position="slave" timeout="20" \ op monitor interval="20" position="grasp" timeout="20 ms ms-drbd-r0 drbd-r0 \ meta interleave="authentic" grasp-max="2" master-nodemax="1" notify="actual" is-managed="actual".
45 OCFS2 Setup.
46 OCFS2 Setup Run the CRM commands to configure the distributed Lock supervisor (DLM) node2:~ # crm configure primitive dlm ocf:pacemaker:controld \ op display screen interval="60" timeout="60 primitive o2cb ocf:ocfs2:o2cb \ op computer screen interval="60" timeout="60 group grp-o2cb dlm o2cb Clone clone-o2cb grp-o2cb meta \ interleave="true". Commit.
47 Create the File system as soon as the supportive items are in region we are able to create the file equipment with this command. mkfs.ocfs2 -T vmstore /dev/drbd_r0 examine the brand new file system through mounting it, writing a file, studying the file from the 2nd node. node1:~ # mount -t ocfs2 /dev/drbd_r0 /statistics/ node1:~ # contact /records/hellotherefromnode1 node2:~ # ls /statistics/ HelloThereFromNode1 misplaced+found
48 Add File device aid Primitive prim-ocfs2-statistics ocf:heartbest:filesystem \ prams equipment= /dev/drbd_r0 listing= /facts/ \ fstype= ocfs2 \ op video display interval= 20 timout= forty birth-lengthen= 10 \ op delivery interval= 0 timeout= 60 \ op stop interval= 0 timeout= 60 clone clone-ocfs prim-ocfs2-records \ meta interleave= authentic Delete base-clone base-group group grp-ocfs2 dlm o2cb prim-ocfs2-records Clone ocfs2-clone grp-ocfs2 meta interleave="actual order drbd-earlier than-ocfs2 inf: ocfs2-clone ms-drbd-r0 \ forty eight clone-o2cb
49 KVM / XEN.
50 Xen visitor resource Agent primitive xen-vm1-vm ocf:heartbeat:xen \ meta enable-migrate="genuine" precedence="4" \ goal-function="begun" \ params xmfile="/etc/xen/vm/xen-vm1" \ op video display interval="60" timeout="240" \ op start interval="0" timeout="120" \ op stop interval="0" timeout="1800" \ op migrate_to interval="0" timeout="1800" \ op migrate_from interval="0" timeout="1800"
51 KVM visitor resource Agent fifty one primitive kvm-guest1 VirtualDomain \ \ \ hypervisor="qemu:///equipment"\ config="/root/guest1.xml"\ meta allow-migrate="true" precedence="four" goal-position="begun" \ op video display interval="60" timeout="240" op birth interval="0" timeout="one hundred twenty" \ op stop interval="0" timeout="1800" \ op migrate_to interval="0" timeout="1800" \
52 Additional Configuration Later types of SBD permit dual shared storage. One project could be to make each xen host an iscsi target for the sbd provider. Then your xen visitors can take advantage of sbd in any other case there is a stonith agent so as to without difficulty spoil a xen guest. As neatly.
53 New in SUSE Linux enterprise.
54 What's New in SUSE Linux business Server 12 complete stack refresh latest edition of products that make up HAE DRBD is still eight.4 SBD is a good deal more reliable Hawk Is favourite cluster config device.
55 Questions? Thanks.
57 Unpublished Work of SUSE LLC. All Rights Reserved. This work is an unpublished work and carries confidential, proprietary and exchange secret suggestions of SUSE LLC. Entry to this work is limited to SUSE personnel who've a need to know to operate initiatives in the scope of their assignments. No part of this work can be practiced, carried out, copied, distributed, revised, modified, translated, abridged, condensed, expanded, amassed, or adapted without the prior written consent of SUSE. Any use or exploitation of this work without authorization may discipline the perpetrator to criminal and civil liability. Widespread Disclaimer This document is not to be construed as a promise via any taking part business to enhance, deliver, or market a product. It is not a dedication to carry any fabric, code, or performance, and should now not be relied upon in making procuring selections. SUSE makes no representations or warranties with recognize to the contents of this doc, and in particular disclaims any specific or implied warranties of merchantability or health for any certain purpose. The construction, unlock, and timing of features or performance described for SUSE products continues to be at the sole discretion of SUSE. Further, SUSE reserves the right to revise this doc and to make adjustments to its content, at any time, without obligation to inform any person or entity of such revisions or changes. All SUSE marks referenced in this presentation are trademarks or registered logos of Novell, Inc. in the United States and other countries. All third-celebration emblems are the property of their respective homeowners.
58 Unpublished Work of SUSE LLC. All Rights Reserved. This work is an unpublished work and carries personal, proprietary and exchange secret counsel of SUSE LLC. Access to this work is limited to SUSE employees who have a need to grasp to function tasks inside the scope of their assignments. No a part of this work can be practiced, carried out, copied, allotted, revised, modified, translated, abridged, condensed, increased, amassed, or adapted devoid of the prior written consent of SUSE. Any use or exploitation of this work devoid of authorization might area the perpetrator to crook and civil liability. Familiar Disclaimer This doc is not to be construed as a promise via any taking part business to improve, deliver, or market a product. It isn't a commitment to carry any cloth, code, or functionality, and may no longer be relied upon in making purchasing selections. SUSE makes no representations or warranties with recognize to the contents of this document, and exceptionally disclaims any categorical or implied warranties of merchantability or fitness for any specific goal. The construction, unlock, and timing of points or functionality described for SUSE items is still on the sole discretion of SUSE. Further, SUSE reserves the correct to revise this document and to make changes to its content, at any time, without obligation to notify any person or entity of such revisions or adjustments. All SUSE marks referenced in this presentation are trademarks or registered trademarks of Novell, Inc. in the U.S. and different countries. All third-party trademarks are the property of their respective homeowners.
No te pierdas el tema anterior: Curso completo sobre trabajo en Linux