Visualization

• Capturing metrics is not enough, it's important to have a place where you can immediately go to and watch the trends of the metrics over time
• We ship Grafana configured by default with
  • Prometheus instance that has all the metrics as a data source
  • Enriched dashboards for different components for exploration of data
• We’ve curated dashboards for analyzing different components of the system
  • API
  • UI
  • MongoDB
  • PostgreSQL
• Almost all the dashboards have some of the standard common panels regarding the compute resource, cpu and ram usage, etc.
  • These panels will most likely be present in almost every component’s dashboard
  • Only RAM Request Utilization and CPU Request Utilization might be missing as in most of the cases these values will be same as the RAM/CPU Limit Utilization as Request and Limit for the resources will be the same

Where can you view the dashboards?

Endpoint

The endpoint can be different in different scenarios

• If ingress enabled
  • <ui-endpoint>/grafana
  • E.g https://staging.dynamofl.com/grafana
• else
  • Domain specific in deployment configuration
  • E.g https://grafana.dynamofl.com

Creds

Configured via the deployment config

• Default username: admin
• Password: get it from the deployment team

Grafana: Entry Point

Click on Dashboards on the left menu panel and you should be able to see the list of the dashboards

Dashboards

Common Panels

Breakdown

Panel Name	Description
Container CPU Request	The CPU Request requested by the resource’s container
Container CPU Limit	The CPU Limit requested by the resource’s container
Container RAM Request	The RAM Request requested by the resource’s container
Container RAM Limit	The RAM Limit requested by the resource’s container
Container Readiness Probe	Whether the container is ready or not (1 means yes, 0 means no)
Container Restarts	Number of times the container has restarted
RAM Request Utilization	% of RAM utilized wrt the Request requested for the container
RAM Limit Utilization	% of RAM utilized wrt the Limit requested for the container
CPU Request Utilization	% of CPU utilized wrt the Request requested for the container
RAM Limit Utilization	% of RAM utilized wrt the Limit requested for the container

API

Common Panels

HTTP Stats

Panel Breakdown

Panel Name	Description
API Health	Health of the API. If 1 that means it’s healthy, else it’ll be 0
Total Requests Rate	Rate of total requests
Average Response Time	Average response time across all endpoints
Internal Server Error Response Rate	Requests with internal server error / Total requests
Probe Error Ratio	Requests with errors of 4xx, 5xx / Total requests
Historic P90 response time by endpoint	P90 response time by different endpoints of the API
HTTP Responses by status code	HTTP responses over time with a given status code
Top HTTP Paths used	Number of HTTP requests by path over time

UI

It only has the common panels

MongoDB Stats

This covers the generic information captured from the kubernetes cluster about the mongodb deployment

Panels Breakdown

Overall Row

• Contains the CPU & RAM Limit Utilization panels Pod Wise Row
• Since MongoDB is deployed as a replicaset you can choose the pod for which you want to view the common pod wise information
• There is MongoDB Pod variable at the top to choose the pod name from

MongoDB Analysis

This is where you can see the panels that contain information obtained by querying the MongoDB database

Panels Breakdown

Panel Name	Description
MongoDB Health	Health of MongoDB. If 1 that means it’s healthy, else it’ll be 0
Session Count	Number of sessions over the course of time
Collection Count (Admin DB)	Total number of collections present in MongoDB for admin database
Active Connections (Admin DB)	Total number of active connections at the moment in MongoDB for admin database
Database Sizes	Size of the databases in MongoDB over the course of time

PostgreSQL Stats

This covers the generic information captured from the kubernetes cluster about the PostgreSQL deployment

PostgreSQL Analysis

This is where you can see the panels that contain information obtained by querying the Postgresql databas

Panels Breakdown

Panel Name	Description
DynamoAI Database size	Total size of the dynamofl database
Database sizes	Size of the different databases in PostgreSQL over time
Size of tables	Size of the tables in the database
Rows in tables	The number of rows in the dynamofl database’s tables
Database Sizes	Size of the databases in MongoDB over the course of time

Pod Generic Stats

• This allows you to view the common panel information mentioned in the overview for any pod in the kubernetes cluster
• You can choose the namespace and then the pod from the list of pods present in that namespace using the select list at the top left corner of the dashboard