table of contents
| PEGASUS-KEG(1) | PEGASUS-KEG(1) |
NAME¶
pegasus-keg - kanonical executable for gridsSYNOPSIS¶
pegasus-keg [-a appname] [-t interval |-T interval] [-l logname]
[-P prefix] [-o fn [..]] [-i fn [..]] [-G sz]
[-C] [-e env [..]] [-p parm [..]]
DESCRIPTION¶
The kanonical executable is a stand-in for regular binaries in a DAG - but not for their arguments. It allows to trace the shape of the execution of a DAG, and thus is an aid to debugging DAG related issues.ARGUMENTS¶
The -e, -i, -o and -p arguments allow lists with arbitrary number of arguments. These options may also occur repeatedly on the command line. The file options may be provided with the special filename - to indicate stdout in append mode for writing, or stdin for reading. The -a, -l , -P , -T and -t arguments should only occur a single time with a single argument.This option allows pegasus-keg to
display a different name as its applications. This mode of operation is useful
in make-believe mode. The default is the basename of argv[0].
-e env [..]
This option names any number of environment
variables, whose value should be reported as part of the data dump. By
default, no environment variables are reported.
-i infile [..]
The pegasus-keg binary can work on any
number of input files. For each output file, every input file will be opened,
and its content copied to the output file. Textual input files are assumed.
Each input line is indented by two spaces. The input file content is bracketed
between an start and end section, see below. By default, pegasus-keg
operates in generator mode.
-l logfile
The logfile is the name of a file to
append atomically the self-info, see below. The atomic write guarantees that
the multi-line information will not interleave with other processes that
simultaneously write to the same file. The default is not to use any log
file.
-o outfile [..]
The pegasus-keg can work on any number
of output files. For each output file, every input file will be opened, and
its content copied to the output file. Textual input files are assumed. Each
input line is indented by two spaces. The input file content is bracketed
between an start and end section, see 2nd example. After all input files are
copied, the data dump from this instance of pegasus-keg is appended to
the output file. Without output files, pegasus-keg operates in data
sink mode.
-G size
If you want pegasus-keg to generate a
lot of output, the generator option will do that for you. Just specify how
much, in bytes, you want. This option is off by default.
-C
This option causes pegasus-keg to list
all environment variables that start with the prefix \_CONDOR The
option is useful, if .B pegasus-keg is run as (part of) a Condor job. This
option is off by default.
-p string [..]
Any number of parameters can be reported,
without being specific on their content. Effectively, these strings are copied
straight from the command line. By default, no extra arguments are
shown.
-P prefix
Each line from every input file is indented
with a prefix string to visually emphasize the provenance of an input files
through multiple instances of pegasus-keg. By default, two spaces are
used as prefix string.
-t interval
The interval is an amount of sleep time that
the pegasus-keg executable is to sleep. This can be used to emulate
light work without straining the pool resources. If used together with the
-T spin option, the sleep interval comes before the spin interval. The
default is no sleep time.
-T interval
The interval is an amount of busy spin time
that the pegasus-keg executable is to simulate intense computation. The
simulation is done by random julia set calculations. This option can be used
to emulate an intense work to strain pool resources. If used together with the
-t sleep option, the sleep interval comes before the spin interval. The
default is no spin time.
RETURN VALUE¶
Execution as planned will return 0. The failure to open an input file will return 1, the failure to open an output file, including the log file, will return with exit code 2.EXAMPLE¶
The example shows the bracketing of an input file, and the copy produced on the output file. For illustration purposes, the output file is connected to stdout :$ date > xx $ pegasus-keg -i xx -p a b c -o - --- start xx ---- Thu May 5 10:55:45 PDT 2011 --- final xx ---- Timestamp Today: 20110505T105552.910-07:00 (1304618152.910;0.000) Applicationname: pegasus-keg [3661M] @ 128.9.xxx.xxx (xxx.isi.edu) Current Workdir: /opt/pegasus/default/bin/pegasus-keg Systemenvironm.: x86_64-Linux 2.6.18-238.9.1.el5 Processor Info.: 4 x Intel(R) Core(TM) i5 CPU 750 @ 2.67GHz @ 2660.068 Load Averages : 0.298 0.135 0.104 Memory Usage MB: 11970 total, 8089 free, 0 shared, 695 buffered Swap Usage MB: 12299 total, 12299 free Filesystem Info: / ext3 62GB total, 20GB avail Filesystem Info: /lfs/balefire ext4 1694GB total, 1485GB avail Filesystem Info: /boot ext2 493MB total, 447MB avail Output Filename: - Input Filenames: xx Other Arguments: a b c
RESTRICTIONS¶
The input file must be textual files. The behaviour with binary files is unspecified.AUTHORS¶
Jens-S. Vöckler <voeckler at isi dot edu>| 05/24/2012 |