1DLM_CONTROLD(8) cluster DLM_CONTROLD(8)
2
6 dlm_controld - daemon that configures dlm according to cluster events
7
10 dlm_controld [OPTIONS]
11
14 The dlm lives in the kernel, and the cluster infrastructure (corosync
15 membership and group management) lives in user space. The dlm in the
16 kernel needs to adjust/recover for certain cluster events. It's the
17 job of dlm_controld to receive these events and reconfigure the kernel
18 dlm as needed. dlm_controld controls and configures the dlm through
19 sysfs and configfs files that are considered dlm-internal interfaces.
20 The cman init script usually starts the dlm_controld daemon.
22
25 Command line options override a corresponding setting in cluster.conf.
26 -D Enable debugging to stderr and don't fork.
29 See also dlm_tool dump in dlm_tool(8).
30 -L Enable debugging to log file.
33 See also logging in cluster.conf(5).
34 -K Enable kernel dlm debugging messages.
37 See also log_debug below.
38 -r num dlm kernel lowcomms protocol, 0 tcp, 1 sctp, 2 detect. 2
41 selects tcp if corosync rrp_mode is "none", otherwise sctp.
42 Default 2.
43 -g num groupd compatibility mode, 0 off, 1 on.
46 Default 0.
47 -f num Enable (1) or disable (0) fencing recovery dependency.
50 Default 1.
51 -q num Enable (1) or disable (0) quorum recovery dependency.
54 Default 0.
55 -d num Enable (1) or disable (0) deadlock detection code.
58 Default 0.
59 -p num Enable (1) or disable (0) plock code for cluster fs.
62 Default 1.
63 -l num Limit the rate of plock operations, 0 for no limit.
66 Default 0.
67 -o num Enable (1) or disable (0) plock ownership.
70 Default 0.
71 -t ms Plock ownership drop resources time (milliseconds).
74 Default 10000.
75 -c num Plock ownership drop resources count.
78 Default 10.
79 -a ms Plock ownership drop resources age (milliseconds).
82 Default 10000.
83 -P Enable plock debugging messages (can produce excessive output).
86 -h Print a help message describing available options, then exit.
89 -V Print program version information, then exit.
92
96 cluster.conf(5) is usually located at /etc/cluster/cluster.conf. It is
97 not read directly. Other cluster components load the contents into
98 memory, and the values are accessed through the libccs library.
99 Configuration options for dlm (kernel) and dlm_controld are added to
101 the <dlm /> section of cluster.conf, within the top level <cluster>
102 section.
103 Kernel options
106 protocol
107 The network protocol can be set to tcp, sctp or detect which
108 selects tcp or sctp based on the corosync rrp_mode configuration
109 (redundant ring protocol). The rrp_mode "none" results in tcp.
110 Default detect.
111 <dlm protocol="detect"/>
113 timewarn
116 After waiting timewarn centiseconds, the dlm will emit a warning
117 via netlink. This only applies to lockspaces created with the
118 DLM_LSFL_TIMEWARN flag, and is used for deadlock detection.
119 Default 500 (5 seconds).
120 <dlm timewarn="500"/>
122 log_debug
125 DLM kernel debug messages can be enabled by setting log_debug to
126 1. Default 0.
127 <dlm log_debug="0"/>
129 clusternode/weight
132 The lock directory weight can be specified one the clusternode
133 lines. Weights would usually be used in the lock server config‐
134 urations shown below instead.
135 <clusternode name="node01" nodeid="1" weight="1"/>
137 Daemon options
140 enable_fencing
141 See command line description.
142 <dlm enable_fencing="1"/>
144 enable_quorum
147 See command line description.
148 <dlm enable_quorum="0"/>
150 enable_deadlk
153 See command line description.
154 <dlm enable_deadlk="0"/>
156 enable_plock
159 See command line description.
160 <dlm enable_plock="1"/>
162 plock_rate_limit
165 See command line description.
166 <dlm plock_rate_limit="0"/>
168 plock_ownership
171 See command line description.
172 <dlm plock_ownership="0"/>
174 drop_resources_time
177 See command line description.
178 <dlm drop_resources_time="10000"/>
180 drop_resources_count
183 See command line description.
184 <dlm drop_resources_count="10"/>
186 drop_resources_age
189 See command line description.
190 <dlm drop_resources_age="10000"/>
192 plock_debug
195 Enable (1) or disable (0) plock debugging messages (can produce
196 excessive output). Default 0.
197 <dlm plock_debug="0"/>
199 Disabling resource directory
203 Lockspaces usually use a resource directory to keep track of which node
204 is the master of each resource. The dlm can operate without the
205 resource directory, though, by statically assigning the master of a
206 resource using a hash of the resource name. To enable, set the per-
207 lockspace nodir option to 1.
208 <dlm>
210 <lockspace name="foo" nodir="1"/>
211 </dlm>
212 Lock-server configuration
215 The nodir setting can be combined with node weights to create a config‐
216 uration where select node(s) are the master of all resources/locks.
217 These master nodes can be viewed as "lock servers" for the other nodes.
218 <dlm>
220 <lockspace name="foo" nodir="1">
221 <master name="node01"/>
222 </lockspace>
223 </dlm>
224 or,
226 <dlm>
228 <lockspace name="foo" nodir="1">
229 <master name="node01"/>
230 <master name="node02"/>
231 </lockspace>
232 </dlm>
233 Lock management will be partitioned among the available masters. There
235 can be any number of masters defined. The designated master nodes will
236 master all resources/locks (according to the resource name hash). When
237 no masters are members of the lockspace, then the nodes revert to the
238 common fully-distributed configuration. Recovery is faster, with lit‐
239 tle disruption, when a non-master node joins/leaves.
240 There is no special mode in the dlm for this lock server configuration,
242 it's just a natural consequence of combining the "nodir" option with
243 node weights. When a lockspace has master nodes defined, the master
244 has a default weight of 1 and all non-master nodes have weight of 0.
245 An explicit non-zero weight can also be assigned to master nodes, e.g.
246 <dlm>
248 <lockspace name="foo" nodir="1">
249 <master name="node01" weight="2"/>
250 <master name="node02" weight="1"/>
251 </lockspace>
252 </dlm>
253 In which case node01 will master 2/3 of the total resources and node2
255 will master the other 1/3.
256
259 dlm_tool(8), fenced(8), cman(5), cluster.conf(5)
260cluster 2009-01-18 DLM_CONTROLD(8)