Process monitoring with additonal metric (instances running) shows in metric browser, but does not show when used in health

[Roland.Clothier]

Hi 

I have used the process monitoring extension and I can see and expand the metrics via the metrics browser on my server it runs on in the controller.   (For example in this case, this process has "running instances" 1 and I can verify this)

I would like to pull this through to a health rule where if running instances < 1 then it will 

The metrics are outlined in /opt/appdynamics/machine-agent/monitors/ProcessMonitor/config.yml

When I use them to create a health rule and use Specify a Relative Metric Path: and use the path copied into the buffer from the metric browser. The "Current health rule evaluation" is greyed out and does not seem to work.

However if I configure the metric in the health rule through to a dashboard it evaluates at true, however if I change the Health rule to trigger an critical condition, it doesnt trigger (as I suspected)

This is also true if I use a metric straight to the dashboard that is not available via the metric tree selection.

Has anybody else ever successfully configured a relative metric to be used in a Health rule or directly as a specify a relative metric path to monitor a process instance count?

Thanks

[Eric.Miller]

Are you trying to report off of 1 particular process or any process running under that custom metrics tier?  If you're trying to get any process, you may need to use a wildcard, to make the path something like:

Application Infrastructure Performance | Tier1 | Custom Metrics | Process Monitor | Linux Processes | * | Running Instances.  It also might help to check the "Set to false if no data" checkbox when designing your rule too.

[Roland.Clothier]

Hi Eric,

Thanks for the reply.  This metric originates from a single machine agent and its one particular process I am trying to get out of the metrics browser in the server view into a dashboard panel for "instances running".  This shows a value of 1 (as it should) when its viewed via the Metrics Browser.

Here is the path but with the servername hidden..

Application Infrastructure Performance|Root|Individual Nodes|myhiddehservername.domain.tld|Hardware Resources|HA Components|Keepalived|Linux Processes|keepalived|Running Instances

Of course then the ProcessMonitor extenstion shows with the server metrics as "HA Components" I want this on a few servers just to monitor HA process instances ( to be sure they are there)

My ProcessMonitor config.yml for for reference

#metricPrefix: "Server|Component:Keepalived|Custom Metrics|Process Monitor|"
metricPrefix: "Hardware Resources|HA Components|Keepalived"
# metricPrefix: "Custom Metrics|Process Monitor|"

# displayName: required - Metrics to be reported under this name in Controller's Metric Browser
# regex/pid/pidFile - process is fetched using this field
instances:
#  - displayName: "machine agent"
#    regex: ".* machineagent.jar"

#  - displayName: "ssh"
#    pid: "1056"

  - displayName: "keepalived"
    pidFile: "/var/run/keepalived.pid"


# Not necessary to modify
linux:
  process: "ps -eo pid,%cpu=CPU%,%mem=Memory%,etime,rsz=RSS,args"

solaris:
  process: "ps -eo pid,pcpu=CPU%, -o pmem=Memory%, -o rss=RSS -o args"

aix:
  process: "ps -eo pid,pcpu=CPU%,pmem=Memory%,rss=RSS,args"


metrics:
  - CPU%:
     multiplier: 1
  - Memory%:
     alias: "Memory%"
  - RSS:
     alias: "Resident Set Size"
  - Running Instances:
     alias: "Running Instances"
  - etime:
     alias: "etime"


# number of concurrent tasks
numberOfThreads: 2

# This is to run this in scheduled mode. In this case, the extension will fetch the data every 300 seconds
# and caches the data. The cached data will be reported to controller every min. This way there is no metric drop.
# Can be used when the data rarely changes or to reduce the load while fetching metrics every minute.
#taskSchedule:
#  numberOfThreads: 1
#  taskDelaySeconds: 300