1NODESET(1) ClusterShell User Manual NODESET(1)
2
3
4
6 nodeset - compute advanced nodeset operations
7
9 nodeset [OPTIONS] [COMMAND] [nodeset1 [OPERATION] nodeset2|...]
10
12 Note: nodeset and cluset are the same command.
13
14 nodeset is an utility command provided with the ClusterShell library
15 which implements some features of ClusterShell's NodeSet and RangeSet
16 Python classes. It provides easy manipulation of 1D or nD-indexed
17 cluster nodes and node groups.
18
19 Also, nodeset is automatically bound to the library node group resolu‐
20 tion mechanism. Thus, it is especially useful to enhance cluster aware
21 administration shell scripts.
22
24 --version
25 show program's version number and exit
26
27 -h, --help
28 show this help message and exit
29
30 -s GROUPSOURCE, --groupsource=GROUPSOURCE
31 optional groups.conf(5) group source to use
32
33 --groupsconf=FILE
34 use alternate config file for groups.conf(5)
35
36 Commands:
37
38 -c, --count
39 show number of nodes in nodeset(s)
40
41 -e, --expand
42 expand nodeset(s) to separate nodes (see also -S SEPA‐
43 RATOR)
44
45 -f, --fold
46 fold nodeset(s) (or separate nodes) into one nodeset
47
48 -l, --list
49 list node groups, list node groups and nodes (-ll) or
50 list node groups, nodes and node count (-lll). When no
51 argument is specified at all, this command will list
52 all node group names found in selected group source
53 (see also -s GROUPSOURCE). If any nodesets are speci‐
54 fied as argument, this command will find node groups
55 these nodes belongs to (individually). Optionally for
56 each group, the fraction of these nodes being member
57 of the group may be displayed (with -ll), and also
58 member count/total group node count (with -lll). If a
59 single hyphen-minus (-) is given as a nodeset, it will
60 be read from standard input.
61
62 -r, --regroup
63 fold nodes using node groups (see -s GROUPSOURCE)
64
65 --groupsources
66 list all active group sources (see groups.conf(5))
67
68 Operations:
69
70 -x SUB_NODES, --exclude=SUB_NODES
71 exclude specified nodeset
72
73 -i AND_NODES, --intersection=AND_NODES
74 calculate nodesets intersection
75
76 -X XOR_NODES, --xor=XOR_NODES
77 calculate symmetric difference between nodesets
78
79 Options:
80
81 -a, --all
82 call external node groups support to display all nodes
83
84 --autostep=AUTOSTEP
85 enable a-b/step style syntax when folding nodesets,
86 value is min node count threshold (integer '4', per‐
87 centage '50%' or 'auto'). If not specified, auto step
88 is disabled (best for compatibility with other cluster
89 tools. Example: autostep=4, "node2 node4 node6" folds
90 in node[2,4,6] but autostep=3, "node2 node4 node6"
91 folds in node[2-6/2].
92
93 -d, --debug
94 output more messages for debugging purpose
95
96 -q, --quiet
97 be quiet, print essential output only
98
99 -R, --rangeset
100 switch to RangeSet instead of NodeSet. Useful when
101 working on numerical cluster ranges, eg. 1,5,18-31
102
103 -G, --groupbase
104 hide group source prefix (always @groupname)
105
106 -S SEPARATOR, --separator=SEPARATOR
107 separator string to use when expanding nodesets
108 (default: ' ')
109
110 -O FORMAT, --output-format=FORMAT
111 output format (default: '%s')
112
113 -I SLICE_RANGESET, --slice=SLICE_RANGESET
114 return sliced off result; examples of SLICE_RANGESET
115 are "0" for simple index selection, or "1-9/2,16" for
116 complex rangeset selection
117
118 --split=MAXSPLIT
119 split result into a number of subsets
120
121 --contiguous
122 split result into contiguous subsets (ie. for nodeset,
123 subsets will contain nodes with same pattern name and
124 a contiguous range of indexes, like foobar[1-100]; for
125 rangeset, subsets with consists in contiguous index
126 ranges)"""
127
128 --axis=RANGESET
129 for nD nodesets, fold along provided axis only. Axis
130 are indexed from 1 to n and can be specified here
131 either using the rangeset syntax, eg. '1', '1-2',
132 '1,3', or by a single negative number meaning that the
133 indices is counted from the end. Because some nodesets
134 may have several different dimensions, axis indices
135 are silently truncated to fall in the allowed range.
136
137 --pick=N
138 pick N node(s) at random in nodeset
139
140 For a short explanation of these options, see -h, --help.
141
142 If a single hyphen-minus (-) is given as a nodeset, it will be read
143 from standard input.
144
146 The nodeset command benefits from ClusterShell NodeSet basic arithmetic
147 addition. This feature extends recognized string patterns by supporting
148 operators matching all Operations seen previously. String patterns are
149 read from left to right, by proceeding any character operators accord‐
150 ingly.
151
152 Supported character operators
153
154 , indicates that the union of both left and right nodeset
155 should be computed before continuing
156
157 ! indicates the difference operation
158
159 & indicates the intersection operation
160
161 ^ indicates the symmetric difference (XOR) operation
162
163 Care should be taken to escape these characters as needed when
164 the shell does not interpret them literally.
165
166 Examples of use of extended patterns
167
168 $ nodeset -f node[0-7],node[8-10]
169 node[0-10]
170
171
172 $ nodeset -f node[0-10]!node[8-10]
173 node[0-7]
174
175
176 $ nodeset -f node[0-10]&node[5-13]
177 node[5-10]
178
179
180 $ nodeset -f node[0-10]^node[5-13]
181 node[0-4,11-13]
182
183
184 Example of advanced usage
185
186 $ nodeset -f @gpu^@slurm:bigmem!@chassis[1-9/2]
187
188 This computes a folded nodeset containing nodes found in group
189 @gpu and @slurm:bigmem, but not in both, minus the nodes found
190 in odd chassis groups from 1 to 9.
191
192 All nodes extension (v1.7+)
193 The @* and @SOURCE:* special notations may be used in extended
194 patterns to represent all nodes (in SOURCE) according to the all
195 external shell command (see groups.conf(5)) and are equivalent
196 to:
197
198 $ nodeset [-s SOURCE] -a -f
199
201 Any wildcard mask found is matched against all nodes from the group
202 source (see groups.conf(5) and the -a/--all option above). * means
203 match zero or more characters of any type; ? means match exactly one
204 character of any type. This can be especially useful for server farms,
205 or when cluster node names differ.
206
207 Say that your group configuration is set to return the following “all
208 nodes”:
209
210 $ nodeset -f -a
211 bckserv[1-2],dbserv[1-4],wwwserv[1-9]
212
213
214 Then, you can use wildcards to select particular nodes, as shown below:
215
216 $ nodeset -f 'www*'
217 wwwserv[1-9]
218
219
220 $ nodeset -f 'www*[1-4]'
221 wwwserv[1-4]
222
223
224 $ nodeset -f '*serv1'
225 bckserv1,dbserv1,wwwserv1
226
227
228 Wildcard masks are resolved prior to extended patterns, but each mask
229 is evaluated as a whole node set operand. In the example below, we
230 select all nodes matching *serv* before removing all nodes matching
231 www*:
232
233 $ nodeset -f '*serv*!www*'
234 bckserv[1-2],dbserv[1-4]
235
236
238 An exit status of zero indicates success of the nodeset command. A
239 non-zero exit status indicates failure.
240
242 Getting the node count
243
244 $ nodeset -c node[0-7,32-159]
245 136
246
247
248 $ nodeset -c node[0-7,32-159] node[160-163]
249 140
250
251
252 $ nodeset -c dc[1-2]n[100-199]
253 200
254
255
256 $ nodeset -c @login
257 4
258
259
260 Folding nodesets
261
262 $ nodeset -f node[0-7,32-159] node[160-163]
263 node[0-7,32-163]
264
265
266 $ echo node3 node6 node1 node2 node7 node5 | nodeset -f
267 node[1-3,5-7]
268
269
270 $ nodeset -f dc1n2 dc2n2 dc1n1 dc2n1
271 dc[1-2]n[1-2]
272
273
274 $ nodeset --axis=1 -f dc1n2 dc2n2 dc1n1 dc2n1
275 dc[1-2]n1,dc[1-2]n2
276
277
278 Expanding nodesets
279
280 $ nodeset -e node[160-163]
281 node160 node161 node162 node163
282
283
284 $ echo 'dc[1-2]n[2-6/2]' | nodeset -e
285 dc1n2 dc1n4 dc1n6 dc2n2 dc2n4 dc2n6
286
287
288 Excluding nodes from nodeset
289
290 $ nodeset -f node[32-159] -x node33
291 node[32,34-159]
292
293
294 Computing nodesets intersection
295
296 $ nodeset -f node[32-159] -i node[0-7,20-21,32,156-159]
297 node[32,156-159]
298
299
300 Computing nodesets symmetric difference (xor)
301
302 $ nodeset -f node[33-159] --xor node[32-33,156-159]
303 node[32,34-155]
304
305
306 Splitting nodes into several nodesets (expanding results)
307
308 $ nodeset --split=3 -e node[1-9]
309 node1 node2 node3
310 node4 node5 node6
311 node7 node8 node9
312
313
314 Splitting non-contiguous nodesets (folding results)
315
316 $ nodeset --contiguous -f node2 node3 node4 node8 node9
317 node[2-4]
318 node[8-9]
319
320
321 $ nodeset --contiguous -f dc[1,3]n[1-2,4-5]
322 dc1n[1-2]
323 dc1n[4-5]
324 dc3n[1-2]
325 dc3n[4-5]
326
327
329 Command syntax has been changed since nodeset command available with
330 ClusterShell v1.1. Operations, like --intersection or -x, are now spec‐
331 ified between nodesets in the command line.
332
333 ClusterShell v1.1:
334
335 $ nodeset -f -x node[3,5-6,9] node[1-9]
336 node[1-2,4,7-8]
337
338
339 ClusterShell v1.2+:
340
341 $ nodeset -f node[1-9] -x node[3,5-6,9]
342 node[1-2,4,7-8]
343
344
345 cluset was added in 1.7.3 to avoid a conflict with xCAT's nodeset com‐
346 mand and also to conform with ClusterShell's "clu*" command nomencla‐
347 ture.
348
350 clubak(1), cluset(1), clush(1), groups.conf(5).
351
352 http://clustershell.readthedocs.org/
353
355 Use the following URL to submit a bug report or feedback:
356 https://github.com/cea-hpc/clustershell/issues
357
359 Stephane Thiell <sthiell@stanford.edu>
360
362 GNU Lesser General Public License version 2.1 or later (LGPLv2.1+)
363
364
365
366
3671.8.1 2018-10-30 NODESET(1)