sge_types.5
NAME
sge_types - Grid Engine type descriptions
DESCRIPTION
The Grid Engine user interface consists of several programs and files.
Some command-line switches and several file attributes are types. The
syntax for these types is explained in this page.
OBJECT TYPES
These types are used for defining Grid Engine configuration:
object_name
An object name is a sequence of up to 512 ASCII printing characters
except SPACE, "/", ":", "'", "\", "[", "]", "{", "}", "|", "(", ")",
"@", "%", "," or the '"' character itself.
calendar_name
A calendar name is the name of a Grid Engine calendar described in
calendar_conf(5).
calendar_name := object_name
ckpt_name
A "ckpt_name" is the name of a Grid Engine checkpointing interface
described in checkpoint(5).
ckpt_name := object_name
complex_name
A complex name is the name of a Grid Engine resource attribute
described in complex(5).
complex_name := object_name
host_identifier
A host identifier can be either a host name or a host group name.
host_identifier := host_name | hostgroup_name
hostgroup_name
A host group name is the name of a Grid Engine host group described in
hostgroup(5). Note, to allow host group names to be distinguished
easily from host names, a "@" prefix is used.
hostgroup_name := @object_name
host_name
A host name is the official name of a host node. Host names with a
domain specification such as "gridmaster.sun.com" are called fully-
qualified host names, whereas host names like "gridmaster" are called
short host names. Note that the install time parameters default_domain
and ignore_fqdn (see bootstrap(5)) affect how Grid Engine deals with
host names in general.
The following host names are generally invalid or reserved:
global, template, all, default, unknown, none. However, it may
sometimes be useful to define a dummy host name of global for
convenient use of qhost(1).
jsv_url
The jsv_url has following format:
jsv_url := jsv_client_url | jsv_server_url
jsv_server_url := [ type ':' ] [ user '@' ] path
jsv_client_url := [ type ':' ] path
type := 'script'
At the moment only the type script is allowed. This means that path is
either the path to a script or to a binary application which will be
used to instantiate a JSV process. The type is optional until other
types are supported by Grid Engine.
Specifying a user is only allowed for server JSVs. Client JSVs will
automatically be started as the submit user, and server JSVs as the
admin user if not otherwise specified.
The path has always to be the absolute path to a binary or application.
memory_specifier
Memory specifiers are positive decimal, hexadecimal or octal integer
constants which may be followed by a multiplier letter. Valid
multiplier letters are k, K, m, M, g, G, t, and T, where k means
multiply the value by 1000, K multiply by 1024, m multiply by
1000x1000, M multiply by 1024x1024, g multiply by 1000x1000x1000, G
multiply by 1024x1024x1024, t multiply by 1000x1000x1000x1000, and T
multiply by 1024x1024x1024x1024. If no multiplier is present, the
value is just counted in bytes. Whether memory values above the 32-bit
limit are representable on 32-bit systems, even for disk space, is
system-dependent.
pe_name
A PE name is the name of a Grid Engine parallel environment described
in sge_pe(5).
pe_name := object_name
project_name
A project name is the name of a Grid Engine project described in
project(5).
project_name := object_name
queue_name
A queue name is the name of a Grid Engine queue described in
queue_conf(5).
queue_name := object_name
time_specifier
A time specifier either consists of a positive decimal, hexadecimal or
octal integer constant, in which case the value is interpreted to be in
seconds, or is built from 3 decimal integer numbers separated by colon
signs, where the first number counts the hours, the second the minutes
and the third the seconds. If a number would be zero it can be left out
but the separating colon must remain (e.g. 1:0:1 = 1::1 means 1 hour
and 1 second).
user_name
A user name can be the name of a login(1) user or of the Grid Engine
user object described in user(5).
user_name := object_name
userset_name
A user set name is the name of a Grid Engine access list or department
described in access_list(5).
userset_name := object_name
date_time
A date_time value must conform to [[CC]YY]MMDDhhmm[.SS], where:
CC denotes the century in 2 digits.
YY denotes the year in 2 digits.
MM denotes the month in 2 digits.
DD denotes the day in 2 digits.
hh denotes the hour in 2 digits.
mm denotes the minute in 2 digits.
ss denotes the seconds in 2 digits (default 00).
If any of the optional date fields are omitted, the
corresponding value of the current date is assumed. If CC is not
specified, a YY of <70 means 20YY.
Use of this option may cause unexpected results if the clocks of
the hosts in the Grid Engine pool are out of sync. Also, the
proper behavior of this option very much depends on the correct
setting of the appropriate timezone, e.g. in the TZ environment
variable (see date(1) for details), when the Grid Engine daemons
sge_qmaster(8) and sge_execd(8) are invoked.
time
A time value must conform to hh:mm:ss, or seconds where:
hh denotes the hour in 2 digits.
mm denotes the minute in 2 digits.
ss denotes the seconds in 2 digits (default 00).
seconds is a number of seconds (used for duration values)
name
A name is an arbitrary string of ASCII printing characters, but may not
contain "/", ":", "@", "\", "*", or "?".
account_name
Identifies the account to which the resource consumption of a job
should be charged.
account_name := name
job_name
A job name is a name as above, with the restriction that it cannot
start with a digit (to avoid ambiguity with a job number in some
contexts).
ar_name
An advance reservation name is a name as above, with the restriction
that it cannot start with a digit (to avoid ambiguity with an AR number
in some contexts).
MATCHING TYPES
These types are used for matching Grid Engine configuration:
expression
A wildcard expression is a regular boolean expression that consists of
one or more patterns joined by boolean operators. When a wildcard
expression is used, the following definition applies:
expression= ["!"] ["("] valExp [")"] [ AND_OR expression ]*
valExp = pattern | expression
AND_OR = "&" | "|"
where:
"!" not operator: negate the following pattern or expression
"&" and operator: logically and with the following expression
"|" or operator: logically or with the following expression
"(" open bracket: begin an inner expression.
")" close bracket: end an inner expression.
"pattern" see the pattern definition that follows
If typed at a shell, the expression itself should be quoted to ensure
that it is not expanded by the shell.
e.g.
"(lx*|sol*)&*64*" any string beginning with either "lx" or
"sol" and containing "64"
"rh_3*&!rh_3.1" any string beginning with "rh_3", except
"rh_3.1"
pattern
When patterns are used the following definitions apply:
"*" matches any character and any number of characters
(between 0 and infinity).
"?" matches any single character.
"." is the character ".". It has no other meaning.
"\" escape character, making the following character match literally;
"\\" matches "\", "\*" matches "*", "\?" matches "?".
"[...]" specifies an array or a range of allowed
characters for one character at a specific position.
Character ranges may be specified using the a-z notation.
The caret symbol (^) is not interpreted as a logical
not; it is interpreted literally.
For more details please see fnmatch(5), glob(7).
A pattern on a shell command line should normally be quoted to avoid it
being interpreted by the shell as a file match.
range
The task range specifier has the form
n[-m[:s]][,n[-m[:s]],...] or n[-m[:s]][ n[-m[:s]] ...]
and thus consists of a comma- or blank-separated list of range
specifiers n[-m[:s]]. The ranges are concatenated to the complete task
id range. Each range may be a single number, a simple range of the form
n-m, or a range with a step size.
wc_ar
The wildcard advance reservation (AR) specification is a placeholder
for AR ids and AR names including AR name patterns. An AR id always
references one AR, while the name and pattern might reference multiple
ARs.
wc_ar := ar_id | ar_name | pattern
wc_ar_list
The wildcard advance reservation (AR) list specification allows
referencing multiple ARs with one command.
wc_ar_list := wc_ar[, wc_ar , ...]
wc_host
A wildcard host specification (wc_host) is a wildcard expression which
might match one or more hosts used in the cluster. The first character
of that string never begins with an at-character ('@'), even if the
expression begins with a wildcard character.
e.g.
* all hosts
a* all host beginning with an 'a'
wc_hostgroup
A wildcard hostgroup specification (wc_hostgroup) is a wildcard
expression which might match one or more hostgroups. The first
character of that string is always an at-character ('@').
More information concerning hostgroups can be found in hostgroup(5)
e.g.
@* all hostgroups in the cluster
@solaris the @solaris hostgroup
wc_job
The wildcard job specification is a placeholder for job ids, and job
names including job name patterns. A job id always references one job,
while the name and pattern might reference multiple jobs.
wc_job := job-id | job_name | pattern
wc_job_range
The wildcard job range specification allows referencing specific array
tasks for one or multiple jobs. The job is referenced via wc_job and in
addition gets a range specifier for the array tasks.
wc_job_range := wc_job [-t range]
If present, the task_range restricts the effect of the qalter etc.
operation to the array job task range specified as a suffix to the job
id. (See the -t option to qsub(1) for further details on array jobs.)
wc_job_list
The wildcard job list specification allows referencing multiple jobs
with one command.
wc_job_list := wc_job[, wc_job, ...]
wc_job_range_list
The wildcard job range list (wc_job_range_list) allows referencing
multiple job ranges with one command. one of the following forms:
wc_job_range_list := wc_job_range[,wc_job_range...]
wc_qdomain
wc_qdomain := wc_cqueue "@" wc_hostgroup
A wildcard expression queue domain specification (wc_qdomain) starts
with a wildcard expression cluster queue name (wc_cqueue) followed by
an at-character '@' and a wildcard expression hostgroup specification
(wc_hostgroup).
A wc_qdomain is used to address a group of queue instances. All queue
instances residing on a host which is part of matching hostgroups will
be addressed. Please note, that wc_hostgroup always begins with an at-
character.
e.g.
*@@* all queue instances whose underlying
host is part of at least one hostgroup
a*@@e* all queue instances beginning with a whose underlying
host is part of at least one hostgroup beginning with e
*@@solaris all queue instances on hosts in
the @solaris hostgroup
wc_cqueue
A wildcard expression cluster queue specification (wc_cqueue) is a
wildcard expression which might match one or more cluster queues used
in the cluster. That string never contains an at-character ('@'), even
if the expression begins with a wildcard character.
e.g.
* all cluster queues
a* all cluster queues beginning with an 'a'
a*&!adam all cluster queues beginning with an 'a', but not adam
wc_qinstance
wc_qinstance := wc_cqueue "@" wc_host
A wildcard expression queue instance specification (wc_qinstance)
starts with a wildcard expression cluster queue name (wc_cqueue)
followed by an at-character '@' and a wildcard expression hostname
(wc_host).
wc_qinstance expressions are used to address a group of queue instances
whose underlying hostname matches the given expression. Please note
that the first character of wc_host never matches the at-character '@'.
e.g.
*@* all queue instances in the cluster
*@b* all queue instances whose
hostname begins with a 'b'
*@b*|c* all queue instances whose
hostname begins with a 'b' or 'c'
wc_queue
wc_queue := wc_cqueue | wc_qdomain | wc_qinstance
A wildcard queue expression (wc_queue) might either be a wildcard
expression, cluster queue specification (wc_cqueue), a wildcard
expression queue domain specification (wc_qdomain), or a wildcard
expression queue instance specification (wc_qinstance).
e.g.
big_*1 cluster queues which begin with
"big_" and end with "1"
big_*&!*1 cluster queues which begin with
"big_", but do not end with "1"
*@fangorn all qinstances residing on host
fangorn
wc_queue_list
wc_queue_list := wc_queue ["," wc_queue "," ...]
Comma-separated list of wc_queue elements.
e.g.
big, medium_*@@sol*, *@fangorn.sun.com
wc_user
A wildcard user name pattern is either a wildcard user name
specification or a full user name.
wc_user := user_name | pattern
wc_user_list
A list of user names.
wc_user_list := wc_user[,wc_user,...]
wc_project
A wildcard project name pattern is either a wildcard project name
specification or a full project name.
wc_project := project | pattern
wc_pe_name
A wildcard parallel environment name pattern is either a wildcard PE
name specification or a full PE name.
wc_pe_name := pe_name | pattern
parallel_env
The parallel_env specification has the format
wc_pe_name n[-[m]]|[-]m,...
specifying a parallel programming environment (PE) to select for a
submitted job or an AR. The range descriptor following the wildcard PE
name specifies the number of slots to allocate, which is usually
equivalent to the total number of parallel processes to be run (for
simple distributed memory jobs) or the number of threads (for shared
memory or mixed distributed/threaded jobs), as implied by the PE
definition. Grid Engine will allocate the appropriate resources, as
available. sge_pe(5) contains information about the definition of PEs.
You can specify a PE name which uses wildcards. Thus the request
"mpi*" will match any parallel environment with a name starting with
the string "mpi". In the case of multiple parallel environments whose
names match a name string, when it is required to select one the one
with the most available slots is chosen.
The range specification is a list of range expressions of the form
"n-m", where n and m are positive, non-zero integers. The form "n" is
equivalent to "n-n". The form "-m" is equivalent to "1-m". The form
"n-" is equivalent to "n-infinity". The range specification is
processed as follows: The largest number of queues requested is checked
first. If enough queues meeting the specified attribute list are
available, all are reserved. If not, the next smaller number of queues
is checked, and so forth.
SEE ALSO
qacct(1), qconf(1), qquota(1), qsub(1), qrsub(1)
COPYRIGHT
See sge_intro(1) for a full statement of rights and permissions.
SGE 8.1.3pre 2011-06-22 SGE_TYPES(5)
Man(1) output converted with
man2html