metal_ping_disco

metal_ping_disco is a "demo application" intended to be used as a mock or stand-in for what a Metal Operator's application could be. It is strictly a lab or demo project, not to be associated with any kind of production setting. When provisioned it will provide a dashboard with a graph of the observed latency between the Equinix Metal host application is hosted on and various pre-defined pingable endpoints.

metal_ping_disco is really two little applications:

• metal_ping_disco.py is meant to be a stateless but long-lived HTTP / browser reachable endpoint that loads data from s3 (collected by data_runner), and graphs it.
• metal_ping_disco_data_runner.py is meant to be a scheduled taks (cronjob) that when run, will load the previous runs dataset (if not found, create) and then ping a collection of endpoints, mangle the data into .csv, and then store that file back to s3.

If this looks useful, you likely really want a real production ready tool such as:

• smokeping
  • There is a Metal specific smokeping document here
• ping-statsd
• ping_exporter for prometheus

Installation

These applications are intended to be run as containerized applications, and should be deployed via docker, podman or containerd.

Both metal_ping_disco and metal_ping_disco_data_runner are hosted on Dockerhub.

Environment Specific Steps:

• Docker
• Podman
• Kubernetes
• Coming soon - systemd

Credentials

The below environment variables need to be set:

export METAL_PING_DISCO_AWS_S3_BUCKET= \
export METAL_PING_DISCO_AWS_ACCESS_KEY_ID= \
export METAL_PING_DISCO_AWS_SECRET_ACCESS_KEY= \
export METAL_PING_DISCO_AWS_S3_ENDPOINT=

These set the s3 endpoint (and provider), bucket name and credentials both applications will use. Note that the s3 provider does not need to be AWS s3 itself, just s3 protocol compliant.

Ping Monitoring

By default, Ping Disco will default to a handful of ICMP reachable endpoints. To specify your own, set the following environment variable:

export METAL_PING_DISCO_PING_ENDPOINTS_RAW="metal_sv15:136.144.55.31,metal_fr2:145.40.93.77,metal_dfw2:139.178.82.17" && \

Where - metal_sv15 will be the name of the graph legend for the edpoint, and 136.144.55.31 is the IP address of the endpoint to ping. DNS names should work fine.

MySQL Monitoring

Optionally, set:

export METAL_PING_DISCO_MYSQL_ENDPOINTS_RAW="haha_db:127.0.0.1:3306:disco_user:disco_Pass:disco_db,haha_db2:127.0.0.2:3306:disco_user:disco_Pass:disco_db"

This will enable tell the container to make a MySQL query and measure the latency for: - 127.0.0.1 on port 3306 with the user disco_user with the password disco_Pass, where the line on the graph and legend will be haha_db, and then all the same for haha_db2 on IP 127.0.0.2. - Values for each mysql instance are colon seperated, mysql instances themselves are comma seperated.

Oracle Monitoring:

Optionally, set:

export METAL_PING_DISCO_ORACLE_ENDPOINTS_RAW="oracle_db:139.178.94.137/xe:1521:oracle_user:oracle_pass,"

This will enable tell the container to make a OracleDB query and measure the latency for: - 139.178.94.137 on port 1521 with the user oracle_user with the password oracle_pass, where the line on the graph and legend will be oracle_db - Values for each oracle instance are colon seperated, oracle instances themselves are comma seperated.