garnet icon indicating copy to clipboard operation
garnet copied to clipboard

Garnet Monitoring Integration in Helm Chart

Open vitalygorohovik93 opened this issue 1 year ago • 3 comments

Feature request type

sample request

Is your feature request related to a problem? Please describe

Integrate monitoring capabilities for Garnet within the existing Helm chart to enhance observability and ensure efficient tracking of system performance.

Describe the solution you'd like

  1. Prometheus Setup: Add Prometheus as an optional dependency in the Helm chart. Configure Prometheus to scrape metrics from Garnet endpoints.

  2. Grafana Dashboards: Provide pre-configured Grafana dashboards for visualizing key metrics. Allow customization of dashboards for user-specific needs. 3.Alerting: Set up default alerting rules in Prometheus. Enable integration with notification systems like Slack or email. 4.Helm Values Configuration: Allow users to enable/disable monitoring through Helm values. Provide configuration options for custom metrics and alert thresholds. 5.Documentation: Update Helm chart documentation with setup instructions. Include examples for custom configurations and troubleshooting tips.

Describe alternatives you've considered

No response

Additional context

Integrating monitoring into the Garnet Helm chart will significantly benefit system administrators by providing real-time insights and alerts, ultimately enhancing system reliability and performance.

vitalygorohovik93 avatar Sep 03 '24 07:09 vitalygorohovik93

@vitalygorohovik93 would this be based on the redis RESP protocol that garnet exposes or something specific to garnet? meaning - is there some existing docker image we could simply add as a another container or do we need to create something new?

Meir017 avatar Sep 03 '24 18:09 Meir017

Grafana/Prometheus integration probably requires https://github.com/microsoft/garnet/issues/379

PaulusParssinen avatar Sep 04 '24 14:09 PaulusParssinen

We can use Grafana OSS with the Redis plugin, which comes with a built-in dashboard. The problem is that some panels don't show data, as comments (like COMMANDSTATS, SLOWLOG) are not yet supported. Other panels worked as expected when I tested them locally with Garnet. @PaulusParssinen, we don't need OpenTelemetry; this works with the RESP protocol. Still better to have OpenTelemetry as well.

Vijay-Nirmal avatar Sep 05 '24 06:09 Vijay-Nirmal