This project was forked from pivotalservices/app-metrics-nozzle.
The original README can be viewed here.
Each Foundation that one wishes to visualize will need an instance of this application pushed to it. Then your instance of vz-metrics-adapter will need to primed (configured) with knowledge about the location of each instance of app-metrics-nozzle.
Edit manifest.yml
and un-comment the following lines:
# API_ENDPOINT: https://api.local.pcfdev.io
# DOPPLER_ENDPOINT: wss://doppler.local.pcfdev.io:443
# FIREHOSE_USER: xxxxx
# FIREHOSE_PASSWORD: xxxxx
Next, set values that are appropirate for your environment:
API_ENDPOINT: https://api.sys.your-pcf.com
DOPPLER_ENDPOINT: wss://doppler.sys.your-pcf.com:443
FIREHOSE_USER: username
FIREHOSE_PASSWORD: password
Now push the application to your PCF/CF.
If one requires more setup guidance, please refer to the original README.md.
The username & password provided requires authorization to both the firehose and the Cloud Controller API (as the same credentials are used to access both services).
The environment information has been removed from the output of app-metrics-nozzle (So that credentials passed into the application via the environment are not leaked to the Internet via this application instance).
-
This project uses a very old version of the firehose2syslog library, and unfortunately it contains a bug that causes it to crash under high message load from the firehose with the error
ERR fatal error: concurrent map read and map write
. Updating to the latest f2s library is non-trivial. We're planning to address this problem in the next iteration of this project, likely by re-writing the core. -
The current behaviour of this project drops all accumulated statistical data when the list of applications is refreshed from the Cloud Controller API (resulting in a drop out of the data). We would like to change this behaviour to use a Ring Buffer pattern so that a rolling history of stats can be maintained between refreshes of the Application list.
As always, PR's welcome!