Ubuntu Manpage: pbs_sched_tcl - pbs Tcl scheduler

Provided by: torque-scheduler_2.4.16+dfsg-1.3ubuntu1.1_amd64

NAME

       pbs_sched_tcl - pbs Tcl scheduler

SYNOPSIS

       pbs_sched  [-a alarm]  [-b file]  [-d home]  [-i file]  [-L logfile]  [-p file]  [-S port]
       [-t file] [-v] [-c file]

DESCRIPTION

       The pbs_sched program runs in conjunction with the PBS  server.   It  queries  the  server
       about  the  state of PBS and communicates with pbs_mom to get information about the status
       of running jobs, memory available etc.  It then makes decisions as to what jobs to run.

       pbs_sched must be executed with root permission.

OPTIONS

-a alarm This specifies the time in seconds to wait for a schedule run to finish.
If a script takes too long to finish, an alarm signal is sent, and the
scheduler is restarted. If a core file does not exist in the current
directory, abort() is called and a core file is generated. The default for
alarm is 180 seconds.

-b file This specifies the "body" file. The file given is read into memory once at
program start or after the program receives a SIGHUP and executed each time
the scheduler is awakened by the server. If this option is not given, the
file "sched_tcl" in the directory PBS_HOME/sched_priv is read for the body
code.

-d home This specifies the PBS home directory, PBS_HOME. The current working
directory of the scheduler is PBS_HOME/sched_priv. If this option is not
given, PBS_HOME defaults to $PBS_SERVER_HOME as defined during the PBS
build procedure.

-i file This specifies the "initialize" file. The file given is executed once
before the main processing loop is entered. If this option is not given,
no initialization code is executed.

-L logfile Specifies an absolute path name of the file to use as the log file. If not
specified, the scheduler will open a file named for the current date in the
PBS_HOME/sched_logs directory (see the -d option).

-p file This specifies the "print" file. Any output from the Tcl code which is
written to standard out or standard error will be written to this file. If
this option is not given, the file used will be
PBS_HOME/sched_priv/sched_out. See the -d option.

-S port This specifies the port to use. If this option is not given, the default
port for the PBS scheduler is used.

-t file This specifies the "terminator" file. If a QUIT command is sent from the
server, this code is executed before the scheduler exits. If this option
is not given, no special termination handling is done.

-v This puts the scheduler into "verbose" mode. Any errors will be shown no
matter what this may be set to, but some "uninteresting" events may be
logged by using this flag. An example is a message each time the server
contacts the scheduler.

-c file Specify a configuration file, see description below. If this is a relative
file name it will be relative to PBS_HOME/sched_priv, see the -d option.
If the -c option is not supplied, pbs_sched will not attempt to open a
configuration file.

The options that specify file names may be absolute or relative. If they are relative,
their root directory will be PBS_HOME/sched_priv.

USAGE

This version of the scheduler requires knowledge of the Tcl language. A set of functions
to communicate with the PBS server and resource monitor have been added to those normally
available with Tcl. All these calls will set the Tcl variable "pbs_errno" to a value to
indicate if an error occured. In all cases, the value "0" means no error. If a call to a
Resource Monitor function is made, any error value will come from the system supplied
errno variable. If the function call communicates with the PBS Server, any error value
will come from the error number returned by the server.

openrm host ?port?
Creates a connection to the PBS Resource Monitor on host using port as the port
number or the standard port for the resource monitor if it is not given. A
connection handle is returned. If the open is successful, this will be a non-
negative integer. If not, an error occurred.

closerm connection
The parameter connection is a handle to a resource monitor which was previously
returned from openrm. This connection is closed. Nothing is returned.

downrm connection
Sends a command to the connected resource monitor to shutdown. Nothing is returned.

configrm connection filename
Sends a command to the connected resource monitor to read the configuration file
given by filename. If this is successful, a "0" is returned, otherwise, "-1" is
returned.

addreq connection request
A resource request is sent to the connected resource monitor. If this is
successful, a "0" is returned, otherwise, "-1" is returned.

getreq connection
One resource request response from the connected resource monitor is returned. If
an error occurred or there are no more responses, an empty string is returned.

allreq request
A resource request is sent to all connected resource monitors. The number of
streams acted upon is returned.

flushreq
All resource requests previously sent to all connected resource monitors are flushed
out to the network. Nothing is returned.

activereq
The connection number of the next stream with something to read is returned. If
there is nothing to read from any of the connections, a negative number is returned.

fullresp flag
Evaluates flag as a boolean value and sets the response mode used by getreq to full
if flag evaluates to "true". The full return from a resource monitor includes the
original request followed by an equal sign followed by the response. The default
situation is only to return the response following the equal sign. If a script
needs to "see" the entire line, this function may be used.

pbsstatserv
The server is sent a status request for information about the server itself. If the
request succeeds, a list with three elements is returned, otherwise an empty string
is returned. The first element is the server's name. The second is a list of
attributes. The third is the "text" associated with the server (usually blank).

pbsstatjob
The server is sent a status request for information about the all jobs resident
within the server. If the request succeeds, a list is returned, otherwise an empty
string is returned. The list contains an entry for each job. Each element is a
list with three elements. The first is the job's jobid. The second is a list of
attributes. The attribute names which specify resources will have a name of the
form "Resource_List:name" where "name" is the resource name. The third is the
"text" associated with the job (usually blank).

pbsstatque
The server is sent a status request for information about all queues resident within
the server. If the request succeeds, a list is returned, otherwise an empty string
is returned. The list contains an entry for each queue. Each element is a list
with three elements. This first is the queue's name. The second is a list of
attributes similar to pbsstatjob. The third is the "text" associated with the queue
(usually blank).

pbsstatnode
The server is sent a status request for information about all nodes defined within
the server. If the request succeeds, a list is returned, otherwise an empty string
is returned. The list contains an entry for each node. Each element is a list with
three elements. This first is the nodes's name. The second is a list of attributes
similar to pbsstatjob. The third is the "text" associated with the node (usually
blank).

pbsselstat
The server is sent a status request for information about the all runnable jobs
resident within the server. If the request succeeds, a list similar to pbsstatjob
is returned, otherwise an empty string is returned.

pbsrunjob jobid ?location?
Run the job given by jobid at the location given by location. If location is not
given, the default location is used. If this is successful, a "0" is returned,
otherwise, "-1" is returned.

pbsasyrunjob jobid ?location?
Run the job given by jobid at the location given by location without waiting for a
positive response that the job has actually started. If location is not given, the
default location is used. If this is successful, a "0" is returned, otherwise, "-1"
is returned.

pbsrerunjob jobid
Re-runs the job given by jobid. If this is successful, a "0" is returned,
otherwise, "-1" is returned.

pbsdeljob jobid
Delete the job given by jobid. If this is successful, a "0" is returned, otherwise,
"-1" is returned.

pbsholdjob jobid
Place a hold on the job given by jobid. If this is successful, a "0" is returned,
otherwise, "-1" is returned.

pbsmovejob jobid ?location?
Move the job given by jobid to the location given by location. If location is not
given, the default location is used. If this is successful, a "0" is returned,
otherwise, "-1" is returned.

pbsqenable queue
Set the "enabled" attribute for the queue given by queue to true. If this is
successful, a "0" is returned, otherwise, "-1" is returned.

pbsqdisable queue
Set the "enabled" attribute for the queue given by queue to false. If this is
successful, a "0" is returned, otherwise, "-1" is returned.

pbsqstart queue
Set the "started" attribute for the queue given by queue to true. If this is
successful, a "0" is returned, otherwise, "-1" is returned.

pbsqstop queue
Set the "started" attribute for the queue given by queue to false. If this is
successful, a "0" is returned, otherwise, "-1" is returned.

pbsalterjob jobid attribute_list
Alter the attributes for a job specified by jobid. The parameter attribute_list is
the list of attributes to be altered. There can be more than one. Each attribute
consists of a list of three elements. The first is the name, the second the
resource and the third is the new value. If the alter is successful, a "0" is
returned, otherwise, "-1" is returned.

pbsrescquery resource_list
Obtain information about the resources specified by resource_list. This will be a
list of strings. If the request succeeds, a list with the same number of elements
as resource_list is returned. Each element in this list will be a list with four
numbers. The numbers specify available, allocated, reserved, and down in that
order.

pbsrescreserve resource_id resource_list
Make (or extend) a reservation for the resources specified by resource_list which
will be given as a list of strings. The parameter resource_id is a number which
provides a unique identifier for a reservation being tracked by the server. If
resource_id is given as "0", a new reservation is created. In this case, a new
identifier is generated and returned by the function. If an old identifier is used,
that same number will be returned. The Tcl variable "pbs_errno" will be set to
indicate the success or failure of the reservation.

pbsrescrelease resource_id
The reservation specified by resource_id is released.

The two following commands are not normally used by the scheduler. They are included here
because there could be a need for a scheduler to contact a server other than the one which
it normally communicates with. Also, these commands are used by the Tcl tools.

pbsconnect ?server?
Make a connection to the named server or the default server if a parameter is not
given. Only one connection to a server is allowed at any one time.

pbsdisconnect
Disconnect from the currently connected server.

The above Tcl functions use PBS interface library calls for communication with the server
and the PBS resource monitor library to communicate with pbs_mom.

datetime ?day? ?time?
The number of arguments used determine the type of date to be calculated. With no
arguments, the current POSIX date is returned. This is an integer in seconds.

With one argument there are two possible formats. The first is a 12 (or more)
character string specifying a complete date in the following format:
YYMMDDhhmmss

All characters must be digits. The year (YY) is given by the first two (or more)
characters and is the number of years since 1900. The month (MM) is the number of
the month [01-12]. The day (DD) is the day of the month [01-32]. The hour (hh) is
the hour of the day [00-23]. The minute (mm) is minutes after the hour [00-59].
The second (ss) is seconds after the minute [00-59]. The POSIX date for the given
date/time is returned.

The second option with one argument is a relative time. The format for this is
HH:MM:SS

With hours (HH), minutes (MM) and seconds (SS) being separated by colons ":". The
number returned in this case will be the number of seconds in the interval
specified, not an absolute POSIX date.

With two arguments a relative date is calculated. The first argument specifies a
day of the week and must be one of the following strings: "Sun", "Mon", "Tue",
"Wed", "Thr", "Fri", or "Sat". The second argument is a relative time as given
above. The POSIX date calculated will be the day of the week given which follows
the current day, and the time given in the second argument. For example, if the
current day was Monday, and the two arguments were "Fri" and "04:30:00", the date
calculated would be the POSIX date for the Friday following the current Monday, at
four-thirty in the morning. If the day specified and the current day are the same,
the current day is used, not the day one week later.

strftime format time
This function calls the POSIX function strftime(). It requires two arguments. The
first is a format string. The format conventions are the same as those for the
POSIX function strftime(). The second argument is POSIX calendar time in second as
returned by datetime. It returns a string based on the format given. This gives
the ability to extract information about a time, or format it for printing.

The Tcl interpreter is started at program initialization and after a reset (the receipt of
a SIGHUP signal). It is not deleted between scheduling runs so variables which are set in
one can be accessed later.

The "initialize" and "terminator" files are run with no supplied connection to the server.
This means that none of the above functions which talk to the server will work unless
pbsconnect is called first. The "body" file is run with a connection to the server
already established.

CONFIGURATION FILE

       A configuration file may be specified with the -c  option.   This  file  may  be  used  to
       specify  the  hosts  (servers)  which  are allowed to connect to pbs_sched.  The hosts are
       specified in the configuration file in a manor identical to that used in  pbs_mom.   There
       is one line per host with the syntax:
       $clienthost   hostname
       where clienthost and hostname are separated by white space.

       Two  host  names  are  always allowed to connection to pbs_sched, "localhost" and the name
       returned to pbs_sched by the system call gethostname().  These names need not be specified
       in the configuration file.

       The  configuration file must be "secure".  It must be owned by a user id and group id less
       than 10 and not be world writable.

FILES

       $PBS_SERVER_HOME/sched_priv
                 the     default     directory     for     configuration     files,     typically
                 (/usr/spool/pbs)/sched_priv.

Signal Handling

       A C based scheduler will handle the following signals:

       SIGHUP The  server  will  close  and reopen its log file and reread the config file if one
              exists.

       SIGALRM
              If the site supplied scheduling module exceeds the time limit, the Alarm will cause
              the scheduler to attempt to core dump and restart itself.

       SIGINT and SIGTERM
              Will result in an orderly shutdown of the scheduler.

       All other signals have the default action installed.

EXIT STATUS

       Upon normal termination, an exit status of zero is returned.