pmnewlog — stop and restart archive logging for PCP performance metrics

Synopsis

$PCP_BINADM_DIR/pmnewlog [-NPsV?] [-a accessfile] [-c configfile] [-C saveconfig] [-n pmnsfile] [-p pid] [other pmlogger options] archive

Description

pmnewlog may be used to stop and restart a running instance of pmlogger(1). This is most useful for managing multiple sets of Performance Co-Pilot (PCP) archive logs. These archive logs record the history of performance metric values that may be “played back” by other PCP tools, and they form the basis of the VCR paradigm and retrospective performance analysis services common to the PCP toolkit.

In normal usage, pmnewlog would be executed by cron(1) in the wee hours to terminate one PCP archive log and start another, i.e. to perform log rotation.

Even more common, would be the execution of pmnewlog from the PCP archive management script pmlogger_daily(1). In this case, direct end-user execution of pmnewlog is most unlikely.

The mandatory argument archive is the base name for the physical files that will constitute the new archive log.

The pmlogger instance to be stopped and restarted must be running on the same system as pmnewlog and is either the primary logger (the default) or the logger with pid as specified by the -p option.

If the -n option is specified, then pmnewlog will use the namespace in the pmnsfile, rather than the default Performance Metrics Name Space (PMNS).

If no -c option is specified, pmnewlog will use pmlc(1) to connect to the running pmlogger(1) and so determine all those metrics and instances that are subject to mandatory logging or advisory on logging, and the associated logging frequencies. This information is used to synthesize a new pmlogger(1) configuration file. If the -n option is specified, it will also be used for these interactions with pmlc(1).

If the -c option is specified, pmlogger(1) will be restarted with configfile as the configuration file. Normally configfile would be the same configuration file used to start pmlogger(1) in the first place, however note that since pmlogger(1) is restarted, any changes to the logging status made using pmlc(1) will be lost, unless these have also been reflected in changes to configfile.

If configfile does not exist, then a search is made in the directory $PCP_VAR_DIR/config/pmlogger for a file of the same name, and if found that file is used, e.g. if config.mumble does not exist in the current directory and the file $PCP_VAR_DIR/config/pmlogger/config.mumble does exist, then -c config.mumble and -c $PCP_VAR_DIR/config/pmlogger/config.mumble are equivalent.

Access controls specifications for the new pmlogger(1) instance may optionally be provided via the -a option. The contents of accessfile should start with the literal token [access] and conform to the syntax of the access controls section as described for pmlogger(1).

The -C option may be used to save the configuration file that pmnewlog passes to the newly launched pmlogger(1).

If the pmlogger(1) instance needs to be started under the control of pmsocks(1) to connect to a pmcd through a firewall, the -s option may be used.

The -V option enables verbose reporting of the activity. By default no output is generated unless some error or warning condition is encountered.

The -N option enables a “show me” mode, where the actions are echoed, but not executed, in the style of “make -n”. Using -N in conjunction with -V maximizes the diagnostic capabilities for debugging.

The other pmlogger options are as described for pmlogger(1). Note that pmnewlog does not support the following options of pmlogger(1).

-h host

pmnewlog determines the host to which the new pmlogger(1) should connect based upon the current host connection for the old pmlogger(1).

-s samples

The new pmlogger(1) is expected to be long running, and the -s option of pmnewlog takes precedence.

-T runtime

The new pmlogger(1) is expected to be long running.

-V version

The new pmlogger will always create the latest version PCP archive format, and the -V option of pmnewlog takes precedence.

-x fd

The launched pmlogger cannot be controlled by pmRecordControl(3).

Options

The available command line options are:

-a file, --access=file

Specify access controls file for the new pmlogger.

-c file, --config=file

Load configuration from file.

-C file, --save=file

Save the configuration of new pmlogger in file.

-n pmnsfile, --namespace=pmnsfile

Load an alternative Performance Metrics Name Space (PMNS(5)) from the file pmnsfile.

-N, --showme

Perform a dry run.

-p PID, --pid=PID

Restart non-primary logger with PID PID.

-P, --primary

Execute as primary logger instance.

-s, --socks

Use pmsocks(1) to connect.

-V, --verbose

Use verbose reporting.

-?, --help

Display usage message and exit.

Examples

The following sh(1) script could be executed by root via cron(1) to start a new set of archive logs for the primary logger each evening. A more complete version of this script may be found in $PCP_BINADM_DIR/pmlogger_daily, and is documented in the manual page for pmlogger_daily(1).

#!/bin/sh
# start new logs for PCP primary logger on this host

# standard place for logs
LOGDIR=$PCP_LOG_DIR/pmlogger/`hostname`

# each new log is named yymmdd.hh.mm
LOGNAME=`date "+%Y%m%d.%H.%M"`

# do it
[ ! -d $LOGDIR ] && mkdir -p $LOGDIR
cd $LOGDIR
$PCP_BINADM_DIR/pmnewlog -l $LOGDIR/pmlogger.log $LOGDIR

Caveats

If no configfile is specified, the method for synthesizing a configuration file using a pmlc(1) connection to the existing pmlogger(1) is, of necessity, incomplete. In particular, for metrics with dynamic underlying instance domains, it is not possible to identify a configuration that logs all instances of a metric all of the time, so rather the synthesized configuration file requests the continued logging of the set of instances that exist at the time pmlogger(1) is interrogated by pmnewlog.

If this situation is a concern, a fixed configuration file should be used, and passed to pmnewlog via the -c option.

Diagnostics

Due to the precious nature of the archive logs, pmnewlog is rather paranoid in its checking and validation, and will try very hard to ensure that an appropriately configured pmlogger(1) can be restarted, before terminating the existing pmlogger(1).

As a consequence of this checking, pmnewlog tends to generate rather verbose error and warning messages.

Files

archive.meta

metadata (metric descriptions, instance domains, etc.) for the archive log

archive.0

initial volume of metrics values (subsequent volumes have suffixes 1, 2, ...)

archive.index

temporal index to support rapid random access to the other files in the archive log

$PCP_BINADM_DIR/pmlogger_daily

sample script to rotate archives for a number of loggers

SaveLogs

if this directory exists within the directory that the archive files will be created by a new pmlogger(1) then the log file (from pmlogger's -l argument) will be linked into the SaveLogs directory with the name archive.log so it can be inspected at a later time. Because the cron-driven PCP archive management scripts run under the uid of the user “pcp”, SaveLogs typically needs to be owned by the user “pcp”.

PCP Environment

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).

See Also

PCPIntro(1), pmcd(1), pmdumplog(1), pmlc(1), pmlogger(1), pmlogger_daily(1), pmsocks(1), pcp.conf(5), pcp.env(5) and PMNS(5).

Referenced By

pmlogger_check(1).

PCP Performance Co-Pilot