By default, CFEngine's monitoring component cf-monitord records performance data about the system. These include process counts, service traffic, load average and CPU utilization and temperature when available. It also records a three year trend summary based any 'shift'-averages.

Custom measurements promises can monitor or log very specific user data through a generic interface. The end-result is to either generate a periodic time series, like the above mentioned values, or to log the results to custom-defined reports.

Promises of type measurement are written just like all other promises within a bundle destined for the agent concerned, in this case monitor. However, it is not necessary to add them to the bundlesequence, because cf-monitord executes all bundles of type monitor.

code
bundle monitor self_watch
{
  measurements:
    # Follow a special process over time
    # using CFEngine's process cache to avoid resampling

    # Example content from /var/cfengine/state/cf_rootprocs
    #USER                             PID         PPID          PGID          %CPU         %MEM        VSZ        NI         RSS    TTY      NLWP STIME     ELAPSED     TIME COMMAND
    #root                             19103       1             19103         0.2          2.1         71716      0          10676  ?           1 18:09       40:13 00:00:06 /var/cfengine/bin/cf-monitord --no-fork

    # match_value:
    #root \s+                         [0-9.]+ \s+ [0-9.]+  \s+  [0-9.]+ \s+   [0-9.]+ \s+  [0-9.]+ \s+ [0-9]+ \s+ [0-9]+ \s+ ([0-9]+) .*"


     "/var/cfengine/state/cf_rootprocs"

        handle => "cf_monitord_RSS",
        stream_type => "file",
        data_type => "int",
        history_type => "weekly",
        units => "kb",
        match_value => proc_value(".*cf-monitord.*",
        "root\s+[0-9.]+\s+[0-9.]+\s+[0-9.]+\s+[0-9.]+\s+[0-9.]+\s+\s+[0-9.]+\s+[0-9.]+\s+([0-9]+).*"),
        comment => "The ammount of memory (RSS or Resident Set Size) cf-monitored is consuming";
}

body match_value proc_value(x,y)
{
  select_line_matching => "$(x)";
  select_multiline_policy => "sum";
  extraction_regex => "$(y)";
}
code
bundle monitor watch_diskspace
{
  measurements:
    # Discover disk device information
    "/bin/df"
      handle => "free_diskspace_watch",
      stream_type => "pipe",
      data_type => "slist",
      history_type => "static",
      units => "device",
      match_value => file_systems;

}

body match_value file_systems
{
  select_line_matching => "/.*";
  extraction_regex => "(.*)";
}

The general pattern of these promises is to decide whether the source of the information is either a file or pipe, determine the data type (integer, string etc.), specify a pattern to match the result in the file stream and then specify what to do with the result afterwards.

History:

  • Custom measurements open sourced in 3.12.0

Architecture

Information gathered by cf-monitord is stored in ${sys.statedir}/env_data to share with other agents like cf-agent and cf-promises which both provide three variables per measurement with different prefixes: value_ for the latest measurement, av_ for the running average and dev_ for the standard deviation of the running average.

The standard measurements and their variable names are listed in mon special variables.

The data over time is stored in ${sys.statedir}/cf_observations.lmdb which can be examined most easily with the cf-check command:

code
cf-check dump /var/cfengine/state/cf_observations.lmdb

By default in the Masterfiles Policy Framework, cf-serverd uses two variables, def.default_data_select_host_monitoring_include and def.default_data_select_policy_hub_monitoring_include to configure which measurements will be included in enterprise reporting.

On the hub side, reports are collected and measurements data is inserted into the MonitoringHG MonitoringMgMeta and MonitoringYrMeta tables of the Enterprise Hub database.

A diagnostic query to run with a Custom Report in Mission Portal.

Select all unique observables in the database:

code
SELECT distinct(observable) FROM monitoringmgmeta;

To see which measures are collected for each host, select all data from this table and then use report result filters to examine a particular host or measure's having data or not.

code
SELECT * FROM monitoringmgmeta;

Measurement data is presented in Mission Portal in the Measurements App and in the Measurements section of the Host info page.

When policy is changed in regards to monitor bundles, both cf-monitord and cf-serverd should be restarted in order to receive the updated policy.

It is possible to configure masterfiles to restart cf-monitord when variables which affect it's configuration are changed.

All measurements historical data is stored in ${sys.statedir}/cf_observations.lmdb. This is where reporting data is pulled from.

${sys.statedir}/env_data is a current snapshot of measurement values which agents (cf-agent and cf-promises) use to generate the special monitor variables.

${sys.statedir}/ts_key is the source for the list of observables/measurements used in the system. This file includes the integer id, name and other meta data in a slightly human readable form, for example:

code
0,users,Users with active processes - including system users,average users per 2.5 mins,0.000,100.000,1

The ts_key file should not be altered.

Note that if a measurement has always had a value of zero it will not be reported and so not available in Mission Portal Measurements or Host info pages.

It is important to specify a promise handle for measurement promises, as the names defined in the handle are used to determine the name of the log file or variable to which data will be reported. Log files are created under WORKDIR/state. Data that have no history type are stored in a special variable context called mon, analogous to the system variables in sys. Thus the values may be used in other promises in the form $(mon.handle).


Attributes

stream_type

Description: The datatype being collected.

Default: pipe

CFEngine treats all input using a stream abstraction. The preferred interface is files, since they can be read without incurring the cost of a process. However pipes from executed commands may also be invoked.

Type: (menu option)

Allowed input range:

code
pipe
file

Example:

code
stream_type => "pipe";

data_type

Description: The datatype being collected.

When CFEngine observes data, such as the attached partitions in the example above, the datatype determines how that data will be handled. Integer and real values, counters etc., are recorded as time-series if the history type is 'weekly', or as single values otherwise. If multiple items are matched by an observation (e.g. several lines in a file match the given regular expression), then these can be made into a list by choosing slist, else the first matching item will be selected.

Type: (menu option)

Allowed input range:

code
counter
int
real
string
slist

Example:

code
"/bin/df"

    handle => "free_disk_watch",
    stream_type => "pipe",

    data_type => "slist",

    history_type => "static",
    units => "device",
    match_value => file_systems,
    action => sample_min(10,15);

history_type

Description: Whether the data can be seen as a time-series or just an isolated value

Type: (menu option)

Allowed input range:

  • scalar

A single value, with compressed statistics is retained. The value of the data is not expected to change much for the lifetime of the daemon (and so will be sampled less often by cf-monitord).

  • static

A synonym for 'scalar'.

  • log

The measured value is logged as an infinite time-series in $(sys.workdir)/state/<handle>_measure.log.

  • weekly

A standard CFEngine two-dimensional time average (over a weekly period) is retained.

Example:

code
"/proc/meminfo"

     handle => "free_memory_watch",
     stream_type => "file",
     data_type => "int",
     history_type => "weekly",
     units => "kB",
     match_value => free_memory;

Notes:

  • Measurements with history type weekly are collected by CFEngine Enterprise reporting.
  • Measurements with history type static are not collected by CFEngine Enterprise reporting.
  • Measurements with history type scalar are not collected by CFEngine Enterprise reporting.
  • Measurements with history type log are not collected by CFEngine Enterprise reporting.

units

Description: The engineering dimensions of this value or a note about its intent used in plots

This is an arbitrary string used in documentation only.

Type: string

Allowed input range: (arbitrary string)

Example:

code
"/var/cfengine/state/cf_rootprocs"

   handle => "monitor_self_watch",
   stream_type => "file",
   data_type => "int",
   history_type => "weekly",
   units => "kB",
   match_value => proc_value(".*cf-monitord.*",

      "root\s+[0-9.]+\s+[0-9.]+\s+[0-9.]+\s+[0-9.]+\s+([0-9]+).*");

match_value

Type: body match_value

See also: Common body attributes

select_line_matching

Description: Regular expression for matching line location

The expression is anchored, meaning it must match a whole line, and not a fragment within a line.

This attribute is mutually exclusive of select_line_number.

Type: string

Allowed input range: .*

Example:

code
# Editing

body location example
{
select_line_matching => "Expression match.* whole line";
}

# Measurement promises

body match_value example
{
select_line_matching => "Expression match.* whole line";
}

select_line_number

Description: Read from the n-th line of the output (fixed format)

This is mutually exclusive of select_line_matching.

Type: int

Allowed input range: 0,99999999999

Example:

code
body match_value find_line
{
select_line_number => "2";
}

extraction_regex

Description: Regular expression that should contain a single back-reference for extracting a value.

A single parenthesized back-reference should be given to lift the value to be measured out of the text stream. The regular expression is unanchored, meaning it may match a partial string

Type: string

Allowed input range: (arbitrary string)

Example:

code
body match_value free_memory
{
select_line_matching => "MemFree:.*";
extraction_regex => "MemFree:\s+([0-9]+).*";
}

track_growing_file

Description: If true, CFEngine remembers the position to which is last read when opening the file, and resets to the start if the file has since been truncated

This option applies only to file based input streams. If this is true, CFEngine treats the file as if it were a log file, growing continuously. Thus the monitor reads all new entries since the last sampling time on each invocation. In this way, the monitor does not count lines in the log file redundantly.

This makes a log pattern promise equivalent to something like tail -f logfile | grep pattern in Unix parlance.

Type: boolean

Example:

code
bundle monitor watch
{
measurements:

   "/home/mark/tmp/file"

         handle => "line_counter",
    stream_type => "file",
      data_type => "counter",
    match_value => scan_log("MYLINE.*"),
   history_type => "log",
         action => sample_rate("0");

}

#

body match_value scan_log(x)
{
select_line_matching => "^$(x)$";
track_growing_file => "true";
}

#

body action sample_rate(x)
{
ifelapsed => "$(x)";
expireafter => "10";
}

select_multiline_policy

Description: Regular expression for matching line location

This option governs how CFEngine handles multiple matching lines in the input stream. It can average or sum values if they are integer or real, or use first or last representative samples. If non-numerical data types are used only the first match is used.

Type: (menu option)

Allowed input range:

code
average
sum
first
last

Example:

code
body match_value myvalue(xxx)
{
 select_line_matching => ".*$(xxx).*";
 extraction_regex => "root\s+\S+\s+\S+\s+\S+\s+\S+\s+\S+\s+\S+\s+\S+\s+(\S+).*";
 select_multiline_policy => "sum";
}

History: Was introduced in 3.4.0 (2012)