gwlmsend man page on HP-UX

Man page or keyword search:  
man Server   10987 pages
apropos Keyword Search (all sections)
Output format
HP-UX logo
[printable version]

gwlmsend(1M)							  gwlmsend(1M)

NAME
       gwlmsend - send metric data to gWLM for use in a custom policy

SYNOPSIS
       gwlmsend [-h] [-w wait_time] metric [value]

AVAILABILITY
       This  command  is  available  only on gWLM managed nodes (HP-UX systems
       where you run gwlmagent) in /opt/gwlm/bin/ to users logged in as root.

DESCRIPTION
       gwlmsend updates values for metrics named in  gWLM's  custom  policies.
       You  define  these  metrics by specifying them in a custom policy. gWLM
       then depends on you to provide values for the metrics so	 it  can  form
       resource	 requests  based on those metric values. For example, gWLM may
       compare a metric value against a target you specify  and	 then  form  a
       request.	 Alternatively, gWLM may simply scale the metric value to form
       its request.

       Use gwlmsend in a script or on the command line	to  provide  either  a
       single  metric  value or a stream of values separated by white space on
       standard input.

       Presumably, you will use a command to collect  or  provide  values  for
       your  metrics. Ensure this command provides a continuous stream of met‐
       rics for your custom policy's use by starting it in a boot script or by
       having cron start it.

       Metric-collection points are used to transfer the values for metrics to
       gwlmagent.  gwlmsend does not create the metric-collection points; this
       is  done	 by  gwlmagent.	 However, gwlmsend can be told to wait for the
       creation of the metric-collection point through a command-line option.

OPTIONS
       -h   Displays usage information and exits. This	option	overrides  all
	    other options.

       -w wait_time
	    Instructs  gwlmsend	 to wait wait_time seconds for the metric-col‐
	    lection point to be created before exiting. The default  wait_time
	    is 5 seconds.

	    Being  able	 to  specify this delay is useful when gWLM management
	    starts with a system boot: You may need to give the gwlmagent dae‐
	    mon more time to create the metric-collection point.

       metric
	    Specifies  that  the  gWLM	metric	with  the name metric is being
	    updated. This argument is required and must match  a  metric  name
	    specified in:

	    + The  gWLM	 graphical interface in HP Systems Insight Manager, in
	      the Metric field for a custom policy

	    or in

	    + The name attribute in a metric tag in an XML  policy  definition
	      used by gwlm

       value
	    Is	the value for the named metric.	 This numeric argument must be
	    the last argument on the command line.  If value is not  provided,
	    white  space-separated metric values are taken from standard input
	    and sent to gwlmagent.  Any	 unitless  integer  or	floating-point
	    number is considered a legal metric value.

	    For	 information on the use of the metric values, see the WARNINGS
	    section below.

RETURN VALUE
       gwlmsend returns exit status 0 if no errors occur, or 1	if  there  are
       errors.

EXAMPLE
       This  example  shows  how to define a metric in a custom policy using a
       portion of a gWLM XML file and gwlmsend to communicate metric values to
       gWLM. (For information on creating complete XML files for gWLM, see the
       gwlmxml(4) manpage.)

       First, in the XML file, we define a custom policy to  keep  the	metric
       db_tx_time at 3.5.

	      <dataStore>
		   <policyDefinition name="db_policy" type="Custom">
			<cpu>
			     <target>3.5</target>
			     <metric name="db_tx_time" response="Inverse">
			     </metric>
			</cpu>
		   </policyDefinition>
	      </dataStore>

       For the metric, you must specify two items:

       + name

       + response

	 Valid	values	for  response are Direct, Inverse, and Scale.  With an
	 Inverse response, a metric value greater than the target results in a
	 CPU  request  larger than the current CPU resource allocation. If the
	 request produces a larger allocation for  the	workload,  the	metric
	 value decreases. So in this example, with an Inverse response and the
	 metric being a DBMS transaction time, we would	 expect	 the  time  to
	 decrease as more resources are allocated to the DBMS workload associ‐
	 ated with this policy. (For  information  on  the  other  values  for
	 response, see gwlmxml(4).)

       Second,	we  use	 gwlmsend  in  either a shell script or on the command
       line.  If the data is coming from a file, say logfile, we  can  forward
       the data to gWLM as follows:

	      tail -f logfile | gwlmsend db_tx_time

       Alternatively,  we  can	forward	 exactly one metric value at a time as
       follows:

	      gwlmsend db_tx_time value

       NOTE
	      Be careful of I/O buffering when feeding data to gwlmsend.   For
	      example, the following line works as expected:

		     tail -f logfile | gwlmsend db_tx_time

	      However,	adding	awk  or	 other	utilities  that buffer I/O can
	      result in data  being  delayed,  or  not	even  sent,  to	 gwlm‐
	      send--depending on when the buffer is flushed:

		     tail -f logfile | awk '{print $1}' | gwlmsend db_tx_time

WARNINGS
   I/O buffering
       As  indicated in the EXAMPLE section above, I/O buffering can delay the
       sending of data.

   Loss of updated values
       If gwlmsend updates the value of a metric more  than  once  in  a  gWLM
       allocation  interval, only the last update before the interval ended is
       used by gWLM in allocating resources.

   Metric values on any change to an SRD
       If you make any change to an SRD, the metric value for a custom	policy
       for a workload based on:

       + An  nPartition	 or  virtual partition is ignored until it is updated.
	 gWLM tries to maintain the partition's current size until the	metric
	 value is updated.

       + A  pset or fss group is effectively reset to 0--potentially resulting
	 in unintended CPU resource movement.  Resend  metric  values  quickly
	 after	making	SRD changes to ensure gWLM is allocating CPU resources
	 as desired.

   Use of outdated values
       gWLM always uses the last value it received for a metric	 to  determine
       CPU allocations for the associated workload. If your command for updat‐
       ing metric values has exited unexpectedly or does not update regularly,
       gWLM  repeatedly	 uses  the  last  value it received--regardless of the
       workload's current behavior.

   Precision to six significant digits
       gwlmsend is precise to only six significant digits. For example, if you
       specify	a  metric  value such as -12345678 with gwlmsend, the value is
       interpreted as -12345700 as the ending 678 is rounded to 700.

DEPENDENCIES
       gwlmsend depends on gWLM to create metric-collection points.  gwlmagent
       is  responsible	for collecting the metric values from a metric-collec‐
       tion point when a custom policy is defined with a metric	 specification
       and used in a deployed SRD.

AUTHOR
       gwlmsend was developed by HP.

FEEDBACK
       If  you	would  like to comment on the current HP gWLM functionality or
       make suggestions for future releases, please send email to:

	      gwlmfeedback@rsn.hp.com

SEE ALSO
       gwlm(1M), gwlmagent(1M), gwlmxml(4), gwlm(5)

								  gwlmsend(1M)
[top]

List of man pages available for HP-UX

Copyright (c) for man pages and the logo by the respective OS vendor.

For those who want to learn more, the polarhome community provides shell access and support.

[legal] [privacy] [GNU] [policy] [cookies] [netiquette] [sponsors] [FAQ]
Tweet
Polarhome, production since 1999.
Member of Polarhome portal.
Based on Fawad Halim's script.
....................................................................
Vote for polarhome
Free Shell Accounts :: the biggest list on the net