* Name of the monitor. The name is appended to the director on which the monitor is configured.
[-s|--stats]
stat[,stat,...]
* One or more statistics to monitor, separated by commas.
Use the monitor stat-list command to display the available statistics.
Optional arguments
[-p|--period]
collection-period
Frequency at which this monitor collects statistics. Valid arguments are an integer followed by:
ms - milliseconds (period is truncated to the nearest second)
s - seconds (Default)
min - minutes
h - hours
0
- Disables automatic polling.
The default period is 30 seconds.
[-d|--director]context-path,
context-path...
* List of one or more comma-separated directors for which to display statistics.
[-t|--targets]context-path,
context-path...
List of one or more comma-separated targets for which to display statistics. Applicable only to statistics that require a target.
[-f|--force]
Forces the creation of the monitor, even if existing monitors are delayed in their polling.
* - argument is positional.
Description
Performance monitoring collects and displays statistics to determine how a port or volume is being used, how much I/O is being processed, CPU usage, and so on.
Metro node collects and displays performance statistics using two user-defined objects:
monitors - Gather the specified statistics.
monitor sinks - Direct the output to the desired destination. Monitor sinks include the console, a file, or a combination of the two.
The monitor defines the automatic polling period, the statistics to be collected, and the output of the format. The monitor sinks define the output destination.
Polling occurs when:
The timer defined by the monitor’s period attribute has expired.
The monitor has at least one sink with the enabled attribute set to true.
Polling is suspended when:
The monitor’s period is set to 0, and/or
All the monitor’s sinks are either removed or their enabled attribute is set to false
Create short-term monitors to diagnose an immediate problem.
Create longer-term monitors for ongoing system management.
About file rotation and timestamps
The log files created by a monitor’s file sink are automatically rotated when they reach a size of 10 MB. The 10MB file is saved as
filename.csv.n where
n is a number 1 - 10, and output is saved in a new file named filename.csv.n+1.
The
.csv files are rotated up to 10 times.
In the following example, a monitor has exceeded 10MB of output. The initial 10MB are stored in
filename.csv.1. Subsequent output is stored in
filename.csv.
service@sms-cluster-1:/var/log/VPlex/cli> ll my-data.csv* -rw-r--r-- 1 service users 2910722 2012-03-06 21:23 my-data.csv -rw-r--r-- 1 service users 10566670 2012-03-06 21:10 my-data.csv.1
If the second file exceeds, 10B, it is saved as
filename.csv.2, and subsequent output is saved in
filename.csv. Up to 10 such rotations, and numbered
.csv files are supported.
When the file sink is removed or the monitor is destroyed, output to the
.csv file stops, and the current
.csv file is time stamped. For example:
service@sms-cluster-1:/var/log/VPlex/cli> ll my-data.csv* -rw-r--r-- 1 service users 10566670 2012-03-06 21:23 my-data.csv.1 -rw-r--r-- 1 service users 5637498 2012-03-06 21:26 my-data.csv_20120306092614973
Examples
Create a simple monitor with the default period, and no targets:
VPlexcli:/monitoring> monitor create --name TestMonitor --director Director-2-1-B --stats director.fe-read,director.fe-write
Successfully created 1 monitor(s) out of 1.
To create a monitor to collect statistics from the director category on /engines/engine1/directors/Director-2-1-B every 10 seconds: