pmgetopt is used to perform command line option parsing for shell
scripts used in the Performance Co-Pilot (PCP toolkit). It is also
used to generate usage messages for those scripts.
The parameters given to pmgetopt take two forms: initially, options
specific to pmgetopt itself are passed in, and terminated using the
-- mechanism. Thereafter, all of the parameters passed into the
shell script should be passed (usually this is simply the "$@"
The available command line options are:
A configuration file in the format described below is passed to
pmconfig using this option. If this option is omitted, then
pmconfig will read its configuration from the standard input
When parsing the calling shell scripts parameters, error and
usage messages will contain the given program name rather than
referring to pmgetopt itself as the source of the error.
A usage message appropriate for the calling shell script to
present as its own can be generated using the option.
Display the usage message for pmgetopt itself and exit.
pmgetopt parses the given parameters, and produces output in a format
suitable for sourcing in the calling shell script. When both short
and long forms of an argument are allowed by the specification,
pmgetopt will always indicate the short form for simpler shell
processing. If arguments are presented that do not match the
configuration, a request for a usage message (-?) will be generated
for the calling script to respond to. Any non-option parameters will
be echoed back to the calling script preceded by the double-hyphen
delimiter. Thus a script should stop handling options when this
delimiter is detected, and begin the handling of any non-option
Unlike with the shell built-in getopt command, variables like $OPTARG
are not set and the calling script will typically employ use of the
shell built-in eval, set and positional shift commands to ensure
option processing occurs correctly.
The configuration format used by pmgetopt is intended to closely
reflect the usage message which would be generated in the presence of
invalid arguments (or the -?, --help option).
There are primarily two types of configuration line - commands and
options. Commands allow metadata to be passed into the option
processing process, and options are the allowable command line
options that the shell script will accept. Command lines are
preceded by the hash character, whereas option lines will always
begin with a hyphen (either single or double). Any other line in the
configuration, which may include usage headers or descriptive text,
has no impact on the option parsing and will be copied unmodified
into the usage message.
The set of commands is: getopt (provide short-argument option
specification manually, if not present this will be generated from
the options presented), usage (provide short one-line summary used at
the head of the usage message, which will be prefixed by the progname
before reporting), and end which informs pmgetopt to stop processing
further commands and options - any subsequent text encountered will
be simply appended to the usage message.
A short-hand notation exists for each of the standard PCP options, as
described in PCPIntro(1). If any of these options (e.g --host)
appears as a single word on any line, it will be transformed into the
appropriate option for the shell script, including all metadata about
that option (whether it accepts an argument, both short and long
option forms, and so on).
Use of the equals symbol ("=") indicates the presence of a required
argument to any option, for both short and long forms. Any non-
standard option must be accompanied by a non-empty description of
As an example, the following is a valid configuration:
# Usage: [options] node...Options:--archive-d, --delay pause between updates for archive replay--host--interval-i=INST, --insts=INST comma-separated metrics instance list-r output raw counters (no rate conversion)--width=N set the width of each column of output--timezone--help
This configuration will produce the following usage message, when run
$ pmgetopt --usage --progname=clusterstat -- "$@"Usage: clusterstat [options] node...Options:-a FILE, --archive=FILEmetrics source is a PCP log archive-d, --delay pause between updates for archive replay-h HOST, --host=HOST metrics source is PMCD on host-t DELTA, --interval=DELTAsampling interval-i INST, --insts=INST comma-separated metrics instance list-r output raw counters (no rate conversion)--width=N set the width of each column of output-Z TZ, --timezone=TZ set reporting timezone-?, --help show this usage message and exit
Several examples of pmgetopt use form part of the PCP toolkit, in
particular the pcp(1) and pmlogmv(1) scripts provide good reference
Environment variables with the prefix PCP_ are used to parameterize
the file and directory names used by PCP. On each installation, the
file /etc/pcp.conf contains the local values for these variables.
The $PCP_CONF variable may be used to specify an alternative
configuration file, as described in pcp.conf(5).
This page is part of the PCP (Performance Co-Pilot) project.
Information about the project can be found at ⟨http://www.pcp.io/⟩.
If you have a bug report for this manual page, send it to
email@example.com. This page was obtained from the project's upstream
Git repository ⟨https://github.com/performancecopilot/pcp.git⟩ on
2020-08-13. (At that time, the date of the most recent commit that
was found in the repository was 2020-08-11.) If you discover any
rendering problems in this HTML version of the page, or you believe
there is a better or more up-to-date source for the page, or you have
corrections or improvements to the information in this COLOPHON
(which is not part of the original manual page), send a mail to
Performance Co-Pilot PCP PMGETOPT(1)