pmdatrace is a Performance Metrics Domain Agent (PMDA) which exports transaction performance metrics from
application processes which use the pcp_trace library described in pmdatrace(3).
A brief description of the pmdatrace command line options follows:
-d It is absolutely crucial that the performance metrics domain number specified here is unique and
consistent. That is, domain should be different for every PMDA on the one host, and the same domain
number should be used for the same PMDA on all hosts.
-l Location of the log file. By default, a log file named trace.log is written in the current
directory of pmcd(1) when pmdatrace is started, i.e. $PCP_LOG_DIR/pmcd. If the log file cannot be
created or is not writable, output is written to the standard error instead.
-A Host-based access control for pmdatrace. access must be either an allow or deny specification,
using either allow:hostspec:maxconns or disallow:hostspec, where `allow' and `disallow' are
keywords, `hostspec' is a host specification conforming to the format used by both pmcd(1) and
pmlogger(1), and `maxconns' is the maximum number of connections allowed from a given `hostspec'.
Using a maximum connections of zero specifies an unlimited number of connections for the
accompanying `hostspec'.
-I Communicate with pcp_trace clients via the given Internet port. This can alternatively be specified
by setting $PCP_TRACE_PORT in the environment to some valid port number (use of the -I option
overrides this). The default port number is 4323.
-Tperiod defines the aggregation period used to compute the recent averages and extrema. Specified as
a time interval using the syntax described in PCPIntro(1) for the common -t PCP argument, e.g. 30seconds or 1min. The default is 60 seconds.
-M User account under which to run the agent. The default is the unprivileged "pcp" account in current
versions of PCP, but in older versions the superuser account ("root") was used by default.
-N Internally, the aggregation period is divided into bucket divisions, and the rolling average is
recomputed every period/bucket seconds. For example, the defaults correspond to -T 60 and -N 12,
which means the average is recomputed every five seconds for a period covering the prior 60 seconds.
-U This option allows the dimension and scale associated with the observation value metric to be
configured. units is a comma-separated string of six integer values, which are the space dimension,
time dimension, count dimension, space scale, time scale, and count scale, respectively. The
default dimension and scale is ``none'', which is equivalent to presenting ``0,0,0,0,0,0'' as the
argument to -U. The units associated with a metric are most easily viewed using the -d (metric
description) option to pminfo(1). The Install script described below steps through this option
quite explicitly, so it is recommended that the Install script be used for building up the units
specification.
Essentially, the exported metrics provide statistics on the time for completion of each transaction, and
an average count of transactions completed and watch points passed over a given time period.