xcat-core/xCAT-client/pods/man3/noderange.3.pod
2008-06-03 14:56:33 +00:00

264 lines
8.8 KiB
Plaintext

=head1 Name
B<noderange> - syntax for compactly expressing a list of node names
=head1 B<Synopsis>
I<Examples:>
node1,node2,node8,node20,group1
node14-node56,node70-node203,group1-group10
node1,node2,node8,node20,node14-node56,node70-node203
node[14-56]
all,-node129-node256,-frame01-frame03
/node.*
^/tmp/nodes
node10+5
10-15,-13
group1@group2
=head1 B<Description>
B<noderange> is a syntax that can be used in most xCAT commands to conve-
niently specify a list of nodes. The result is that the command will
be applied to a range of nodes, often in parallel.
B<noderange> is a comma-separated list. Each token (text between commas)
in the list can be any of the forms listed below:
Individual node or group:
I<node01>
I<group1>
A range of nodes or groups:
I<node01-node10> (equivalent to: node01,node02,node03,...node10)
I<group1-group3> (equivalent to: group1,group2,group3)
A regular expression match of nodes or groups:
I</node[345].*> (will match any nodes that start with node3, node4, or
node5)
I</group[12].*> (will match any groups that start with group1 or group2)
An incremented range of nodes:
I<node10+3> (equivalent to: node10,node11,node12,node13)
The full path of a file containing noderanges of nodes or groups:
I<^/tmp/nodelist>
A node shorthand range of nodes:
I<10-20> (if prefix is `node', equivalent to:
node10,node11,node12,...node20)
I<10+3> (if prefix is `node', equivalent to: node10,node11,node12,node13)
Or any combination:
I<node01-node30,node40,^/tmp/nodes,/node[13].*,2-10,node50+5>
Any individual B<noderange> may be prefixed with an exclusion operator
(default -) with the exception of the file operator (default ^).
The intersection operator @ calculates the intersection of the left and
right sides:
I<group1@group2> (will result in the list of nodes group1 and group2 have
in common)
Any combination or multiple combinations of inclusive and exclusive
ranges of nodes and groups is legal. There is no precedence implied in
the order of the arguments. Exclusive ranges have precedence over
inclusive.
Nodes have precedence over groups. If a node range match is made then
no group range match will be attempted.
All node names are validated against the nodelist table. Invalid nodes
are ignored and return nothing.
All group names are validated against the nodelist and nodetype table.
Invalid groups are ignored and return nothing.
Throughout this man page the term I<xCAT> I<Node> I<Format> will be used.
I<xCAT> I<Node> I<Format> is defined by the following regex:
^([A-Za-z-]+)([0-9]+)(([A-Za-z-]+[A-Za-z0-9-]*)*)
In plain english a node or group is in I<xCAT> I<Node> I<Format> if starting
from the begining there are one or more alpha characters of any case
and any number of - in any combination, followed by one or more num-
bers, then optionally followed by one alpha character of any case or
the - followed by any combination of case mixed alphanumerics and the
-.
B<noderange> supports node/group names in I<any> format. I<xCAT> I<Node> I<Format> is
B<not> required, however some node range methods used to determine range
will not be used. E.g. If using a B<noderange> of I<node1a-node9a> with a
B<nodelist> table only listing I<node1a> through I<node5a>, B<noderange> will enu-
merate then validate and return a proper range. If using a node range
of I<aa-az> with a B<nodelist> table only listing I<aa> through I<ay,> B<noderange>
will fail to return any values.
Example of I<xCAT> I<Node> I<Format> node/group names:
B< nodename prefix number suffix>
node1 node 1
node001 node 001
node-001 node- 001
node-foo-001-bar node-foo- 001 -bar
node-foo-1bar node-foo- 1 bar
foo1bar2 foo 1 bar2
rack01unit34 rack 01 unit34
unit34rack01 unit 34 rack01
pos0134 pos 0134
Example of non-I<xCAT> I<Node> I<Format> node/group names, but still valid:
aa
yellow
red
12foo
The supported B<noderange> syntaxes are checked for, and processed, in a
specific order. First B<noderange> checks for the multiple range operator
(default ,). Each range is also processed by B<noderange>.
Next B<noderange> checks for the file operator (default ^). If the file
exists (must be a full path name) each line will be processed as a
B<noderange>. Lines starting with I<#> or the file operator (default ^) are
ignored. Only one B<noderange> per line is read. All characters are
ignored after the first white space.
e.g.
^/tmp/nodes
where
cat /tmp/nodes outputs:
#my node list (this line ignored)
^/tmp/foo #ignored
node01 #node comment
node02
node03
node10-node20
/group[456].*
-node50
Next B<noderange> checks for the exclusion operator (default -) then con-
tinues. This operator supports nodes and groups. B<noderange> will not
confuse the exclusion or range operators with the - character in names.
Next B<noderange> checks for a numeric only range (e.g. 10-20, 5+3, or
just 10), then uses I<$XCAT>B<_>I<NODE>B<_>I<PREFIX> (default is `node') and
I<$XCAT>B<_>I<NODE>B<_>I<SUFFIX> (optional) as the defaults to complete the node
names. B<nodeRange> tries to be intelligent about detecting padding, so
you can specify `node001-node200' and it will add the proper number of
zeroes to make all numbers 3 digits. Noderange shorthand supports
nodes only. Noderange shorthand can be mixed with all other operators
except regex. i.e. exclusion, increment, range, and file may be used.
Next B<noderange> checks for the for the regular expression operator
(default /). Regular expressions offer the most flexibility. If you
are interested in learning regex read the book I<Mastering> I<Regular>
I<Expressions>. This operator supports nodes and groups.
Next B<noderange> checks for the increment range operator (default +).
Increment range operator noderanges are in the format:
I<valid>B<_>I<node>B<_>I<name>B<+>I<number>B<_>I<of>B<_>I<sequential>B<_>I<nodes>
e.g.
I<node10+5>
would yield node10 plus the next I<5> nodes.
This action is performed using two different methods. If
I<valid>B<_>I<node>B<_>I<name> is in I<xCAT> I<Node> I<Format> then the range is enumerated to
I<number>B<_>I<of>B<_>I<sequential>B<_>I<nodes>. If not in I<xCAT> I<Node> I<Format> then a sorted
B<nodelist> table is used to return the node range. This operator sup-
ports nodes only.
Next B<noderange> checks for a single node name or group name.
Next B<noderange> checks for the range operator (default -). Ranges are
performed first by validating that both the start and end nodes or
groups defining the range exist and if so the range is returned based
on the content of the B<nodelist> and B<nodetype> tables. If the start and
end nodes or groups defined in the range do not exist, I<and> if both are
in I<xCAT> I<Node> I<Format>, I<and> if both the prefix and suffix match, then the
range is enumerated and each node/group validated. Only valid
nodes/groups will be returned. B<noderange> will not confuse the exclu-
sion or range operators with the - character in names.
Last B<noderange> returns nothing if no match can be found.
B<noderange> uses the smallest integer to determine padding. e.g.
I<node1-node10> will generate a list of nodes with numbers 1, 2, 3, 4, 5,
6, 7, 8, 9, 10. I<node001-node010> will generate a list of nodes with
numbers 001, 002, 003, 004, 005, 006, 007, 008, 009, 010.
Multiple instances of a node name are treated as one instance. e.g.
I<node1-node10,node4,node4,node4> will generate a list of nodes numbered 1
through 10, the number 4 will only be listed once.
=head1 B<Examples>
I<all,-node5-node10>
Generates a list of all nodes (assuming all is a group) listed in the
B<nodelist> table less node5 through node10.
I<node1-node10,-node3-node5,node4>
Generates a list of nodes 1 through 10 less nodes 3,4,5. Note that
node4 is listed twice, first in the range and then at the end. Because
exclusion has precedence node4 will be excluded.
I<node1-node10,-node3,-node5>
Generates a list of nodes 1 through 10 less nodes 3 and 5.
I<-node17-node32,all>
Generates a list of all (assuming `all' is a group) nodes in the
B<nodelist> table less 17 through 32.
I<node1-node128,user1-user4>
Generates a list of nodes 1 through 128, and user nodes 1 through 4.
I<all,-rack1-rack3,-node100-node200,node150,-storage>
Generates a list of all nodes (assuming `all' is a group), less nodes
in groups rack1 through rack3 (assuming groups rack1, rack2, and rack3
are defined), less nodes 100 through 200, less nodes in the storage
group. Note that node150 is listed but is excluded.
I</node[23].*>
Generates a list of nodes matching the regex I<node[23].*>. That is all
nodes that start with node2 or node3 and end in anything or nothing.
E.g. node2, node3, node20, node30, node21234 all match.
=head1 B<SEE ALSO>
nodels(1)