Debian Codegraph @

UG9/en/rhcs.adoc

raw
1 [[ch-rhcs]]
2 == Integrating DRBD with Red Hat Cluster
3
4 indexterm:[Red Hat Cluster]This chapter describes using DRBD as
5 replicated storage for Red Hat Cluster high availability
6 clusters.
7
8 NOTE: This guide uses the unofficial term _Red Hat Cluster_ to refer
9 to a product that has had multiple official product names over its
10 history, including _Red Hat Cluster Suite_ and _Red Hat Enterprise
11 Linux High Availability Add-On_.
12
13 [[s-rhcs-primer]]
14 === Red Hat Cluster background information
15
16 [[s-rhcs-fencing]]
17 ==== Fencing
18
19 Red Hat Cluster, originally designed primarily for shared
20 storage clusters, relies on node fencing to prevent concurrent,
21 uncoordinated access to shared resources. The Red Hat Cluster
22 fencing infrastructure relies on the fencing daemon `fenced`, and
23 fencing agents implemented as shell scripts.
24
25 Even though DRBD-based clusters utilize no shared storage resources
26 and thus fencing is not strictly required from DRBD's standpoint, Red
27 Hat Cluster Suite still requires fencing even in DRBD-based
28 configurations.
29
30 [[s-rhcs-rgmanager]]
31 ==== The Resource Group Manager
32
33 The resource group manager (`rgmanager`, alternatively `clurgmgr`) is
34 akin to Pacemaker. It serves as the cluster management suite's primary
35 interface with the applications it is configured to manage.
36
37 [[s-rhcs-resources]]
38 ===== Red Hat Cluster resources
39
40 indexterm:[Red Hat Cluster]A single highly available
41 application, filesystem, IP address and the like is referred to as a
42 _resource_ in Red Hat Cluster terminology.
43
44 Where resources depend on each other -- such as, for example, an NFS
45 export depending on a filesystem being mounted -- they form a
46 _resource tree_, a form of nesting resources inside another. Resources
47 in inner levels of nesting may inherit parameters from resources in
48 outer nesting levels. The concept of resource trees is absent in
49 Pacemaker.
50
51 [[s-rhcs-services]]
52 ===== Red Hat Cluster services
53
54 indexterm:[Red Hat Cluster]Where resources form a co-dependent
55 collection, that collection is called a _service_. This is different
56 from Pacemaker, where such a collection is referred to as a _resource
57 group_.
58
59 [[s-rhcs-resource-agents]]
60 ===== rgmanager resource agents
61
62 The resource agents invoked by `rgmanager` are similar to those used by
63 Pacemaker, in the sense that they utilize the same shell-based
64 API as defined in the Open Cluster Framework (OCF), although Pacemaker
65 utilizes some extensions not defined in the framework. Thus in theory,
66 the resource agents are largely interchangeable between Red Hat
67 Cluster Suite and Pacemaker -- in practice however, the two cluster
68 management suites use different resource agents even for similar or
69 identical tasks.
70
71 Red Hat Cluster resource agents install into the `/usr/share/cluster/`
72 directory. Unlike Pacemaker OCF resource agents which are by
73 convention self-contained, some Red Hat Cluster resource agents are
74 split into a `.sh` file containing the actual shell code, and a
75 `.metadata` file containing XML resource agent metadata.
76
77 DRBD includes a Red Hat Cluster resource agent. It installs into the
78 customary directory as `drbd.sh` and `drbd.metadata`.
79
80 [[s-rhcs-config]]
81 === Red Hat Cluster configuration
82
83 This section outlines the configuration steps necessary to get Red Hat
84 Cluster running. Preparing your cluster configuration is fairly
85 straightforward; all a DRBD-based Red Hat Cluster requires are two
86 participating nodes (referred to as _Cluster Members_ in Red Hat's
87 documentation) and a fencing device.
88
89 NOTE: For more information about configuring Red Hat clusters, see
90 http://www.redhat.com/docs/manuals/csgfs/[Red Hat's documentation on
91 the Red Hat Cluster and GFS.]
92
93
94 [[s-rhcs-cluster-conf]]
95 ==== The `cluster.conf` file
96
97 RHEL clusters keep their configuration in a single configuration file,
98 indexterm:[Red Hat Cluster]indexterm:[cluster.conf (Red Hat Cluster
99 configuration file)]`/etc/cluster/cluster.conf`. You may manage your
100 cluster configuration in the following ways:
101
102 .Editing the configuration file directly
103 This is the most straightforward method. It has no prerequisites other
104 than having a text editor available.
105
106 .Using the `system-config-cluster` GUI
107 This is a GUI application written in Python using Glade. It requires
108 the availability of an X display (either directly on a server console,
109 or tunneled via SSH).
110
111 .Using the Conga web-based management infrastructure
112 The Conga infrastructure consists of a node agent ( `ricci`)
113 communicating with the local cluster manager, cluster resource
114 manager, and cluster LVM daemon, and an administration web application
115 ( `luci`) which may be used to configure the cluster infrastructure
116 using a simple web browser.
117
118
119 [[s-rhcs-failover-clusters]]
120 === Using DRBD in Red Hat Cluster fail-over clusters
121
122 NOTE: This section deals exclusively with setting up DRBD for Red Hat
123 Cluster fail-over clusters not involving GFS. For GFS (and GFS2)
124 configurations, please see <<ch-gfs>>.
125
126 This section, like <<ch-pacemaker>>, assumes you are about to
127 configure a highly available MySQL database with the following
128 configuration parameters:
129
130 * The DRBD resources to be used as your database storage area is named
131 `mysql`, and it manages the device `/dev/drbd0`.
132
133 * The DRBD device holds an ext3 filesystem which is to be mounted to
134 `/var/lib/mysql` (the default MySQL data directory).
135
136 * The MySQL database is to utilize that filesystem, and listen on a
137 dedicated cluster IP address, 192.168.42.1.
138
139
140 [[s-rhcs-example-cluster-conf]]
141 ==== Setting up your cluster configuration
142
143 To configure your highly available MySQL database, create or modify
144 your `/etc/cluster/cluster.conf` file to contain the following
145 configuration items.
146
147 To do that, open `/etc/cluster/cluster.conf` with your preferred text
148 editing application. Then, include the following items in your
149 resource configuration:
150
151 [source,xml]
152 ----------------------------
153 <rm>
154 <resources />
155 <service autostart="1" name="mysql">
156 <drbd name="drbd-mysql" resource="mysql">
157 <fs device="/dev/drbd/by-res/mysql/0"
158 mountpoint="/var/lib/mysql"
159 fstype="ext3"
160 name="mysql"
161 options="noatime"/>
162 </drbd>
163 <ip address="192.168.42.1" monitor_link="1"/>
164 <mysql config_file="/etc/my.cnf"
165 listen_address="192.168.42.1"
166 name="mysqld"/>
167 </service>
168 </rm>
169 ----------------------------
170
171 NOTE: This example assumes a single-volume resource.
172
173 Nesting resource references inside one another in `<service/>` is the
174 Red Hat Cluster way of expressing resource dependencies.
175
176 Be sure to increment the `config_version` attribute, found on the root
177 `<cluster>` element, after you have completed your
178 configuration. Then, issue the following commands to commit your
179 changes to the running cluster configuration:
180
181 ----------------------------
182 # ccs_tool update /etc/cluster/cluster.conf
183 # cman_tool version -r <version>
184 ----------------------------
185
186 In the second command, be sure to replace _<version>_ with the new
187 cluster configuration version number.
188
189 NOTE: Both the `system-config-cluster` GUI configuration utility and
190 the Conga web based cluster management infrastructure will complain
191 about your cluster configuration after including the `drbd` resource
192 agent in your `cluster.conf` file. This is due to the design of the
193 Python cluster management wrappers provided by these two applications
194 which does not expect third party extensions to the cluster
195 infrastructure.
196
197 Thus, when you utilize the `drbd` resource agent in cluster
198 configurations, it is not recommended to utilize
199 `system-config-cluster` nor Conga for cluster configuration
200 purposes. Using either of these tools to only monitor the cluster's
201 status, however, is expected to work fine.